You are in: Home » History of the Unit
MEMORY
General Notes. This material has been scanned from the original typescript. While we have done our best to remove errors, some may well remain. You can access other parts of this particular Progress Report either from the menu at the bottom of this entry, by using your browser's back function, by navigating back to the Unit History Timeline, or by accessing the relevant section of the electronic archive. Reference for this report are indexed by number and these can be found in a dedicated section also accessible from the menu at the bottom of this entry.
Andrade 3.75, A Baddeley 4.0, Bekerian 5.0, Fulcher 0.92, G Houghton 2.0, Maylor 1.35, Murre 2.12, Nimmo-Smith 0.75, Norris 2.0, Page 0.84, Papagno 1.67, K Patterson 1.0, Sellen 0.73, Shanks 3.25, A Young 0.60, Wilson 1.0, Boden (ASO) 0.25, Dennett (HSO) 5.0, Evans (HSO) 0.83
Total Person Years: Scientists 31.0; Research Support 6..1
Abstract
Objectives
All human cognitive activity depends on our ability to store and retrieve information in memory. In this programme we attempt to understand both the basic processes of human memory and the interaction between memory and other cognitive processes. These goals are pursued by laboratory studies of both normal subjects and subjects with acquired memory disorders, and also by studying behaviour in more complex naturalistic settings.
Scientific progress and achievements over the past five years
The Working Memory model has become an influential framework for the study of memory. Extensive work on developing and extending the scope of the model has continued here at the APU. One of the more significant discoveries stemming from research in the Working Memory framework has been that poor performance in repeating non-words correlates highly with delayed language development. This has led to the development of a non-word repetition test to predict problems in linguistic development. To complement the empirical research on working memory, we have developed computational models of serial recall which have so far proved successful in accounting for a wide range of data from serial recall paradigms. The Competitive Queueing model of serial ordering has been a major influence on other workers in the area.
More naturalistic studies of the manner in which memories change over time have led to the production of important practical guidelines for procedures such as medical interviews and evidential interviews with victims of crime.
In work on the cognitive effects of ageing, we have demonstrated declines that are not explained by traditional accounts based on single-factor models such as proportionate slowing. Disproportionate effects include forgetting rates in prospective memory tasks and retrieval blocks in naming tasks.
Progress has also been made in the study of neuropsychological disorders of memory. This work has shown that prosopagnosia (a deficit in recognising familiar faces following brain insult) can be due to a selective impairment of face memory rather than a more general perceptual deficit. Studies of the deterioration of semantic memory, in both generalised dementia of the Alzheimer's type and more focal "semantic" dementia, address questions about the neuroanatomical basis of semantic memory and also its functional organisation. For example, two detailed case studies of semantic dementia have documented progressive deterioration in specific semantic information with preserved higher-level superordinate knowledge, a pattern of data interpretable as reflecting a loss of specific features from a distributed network of knowledge.
Future plans for the next five years
Future research on memory will continue our present strategy of combining studies of both normal individuals and individuals with acquired memory disorders. Both laboratory studies and work in more naturalistic settings will emphasise the importance of memory as a component in the performance of other cognitive tasks. Much of the work, particularly in the areas of short-term memory and amnesia, will make increasing use of computational modelling and simulation to develop precise formal models in an attempt to gain a more detailed understanding of basic memory mechanisms.
Neuropsychological investigations will continue to be directed primarily towards the topics of semantic memory and face recognition. Patients with Alzheimer's disease will be examined to determine correlation between degree of semantic memory impairment and structural/functional abnormalities in brain regions assumed to be critical for semantic memory. Patients with semantic dementia will be studied to shed light on the distinction between episodic and semantic memory. Disorders of face recognition will focus on retrograde and anterograde components of face memory, and their relation to semantic memory for people.
Implications for improving health, health care and wealth creation
There is currently great concern over the reading abilities of young children. Our work on phonological memory and reading has led to the development of simple predictive tests which can help to identify those children likely to have difficulty in learning to read, and thereby help to optimise allocation of teaching resources.
Studies of memory under anaesthesia are part of the development of a new technique to monitor the depth of anaesthesia which may significantly minimise suffering during and following surgical procedures.
Effective interviewing procedures have an important role to play in ensuring the efficient operation of the criminal justice system. Our research in long-term memory has produced new guidelines for interviewing victims of crime.
Much of our work in memory has important implications for care of the aged. With an increasingly ageing population, it is vital that we have a better understanding of the cognitive limitations that accompany old age. The problems associated with Alzheimer's disease are an area of particular concern. Tests developed to assess the severity and course of the disease will be increasingly valuable as new methods of treatment develop.
WORKING MEMORY AND LONG-TERM MEMORY (Andrade, Baddeley, Emslie, Papagno)
A. Working Memory
Working memory refers to the temporary storage system assumed to be responsible for the maintenance of information necessary for complex skills such as learning, reasoning and comprehension. A tripartite model of working memory was proposed by Baddeley and Hitch some 20 years ago, and since that time has developed gradually. In recent years the model has had a major impact on this area of research, with invitations to Baddeley to give plenary addresses to the Cognitive Science Society and the European Neuroscience Society, and to publish review papers in a wide range of journals, including Science (3.12, 3.13, 3.14, 3.103). The model assumes three subsystems: the phonological loop holds and manipulates linguistic and auditory information; the visuo-spatial sketchpad is the visual counterpart of the phonological loop; both are assumed to be controlled by the central executive, a limited capacity attentional system (3.112).
A1. The Phonological Loop: The previous Progress Report described a patient with a deficit in this system who had specific problems in new phonological learning, such as that required to learn the vocabulary of a foreign language. While we have not had access to another single case as pure and dramatic as that in our original study, two milder cases of developmental deficit have allowed the original findings to be replicated (3.16, 3.27), while other single cases have elucidated the relationship between the phonological loop and comprehension (3.90), and between phonological processing and memory (3.28).
A major component of our recent work in this area has been designed to test the hypothesis that the phonological loop has evolved as a system for the acquisition of language. Much of this work has been carried out in collaboration with Susan Gathercole, who moved from the APU to the University of Lancaster, and subsequently to the University of Bristol. We showed that children with normal nonverbal intelligence, but delayed language development, had a marked deficit in the capacity to hear and repeat back an unfamiliar pseudoword (3.43). On the basis of this result, we developed the nonword repetition test, which we showed to be the best available predictor of vocabulary acquisition in children (3.44, 3.45, 3.50, 3.132). A longitudinal study using cross-lagged correlation indicated that it is good nonword repetition which leads to good vocabulary acquisition, rather than the reverse, in four-year-old children, although as children get older the pattern becomes more bi-directional, with existing vocabulary providing a model for remembering novel (non)words (3.44). More detailed analysis of the test suggests two processes (3.49). One is based on items that resemble English words in structure, presumably reflecting the influence of prior vocabulary learning. The second is based on items that differ from English and hence presumably impose a greater load on phonological short-term storage. Performance on these latter items gives the best prediction of subsequent language development (3.48).
The relationship between nonword repetition and subsequent reading performance is more complex, with the association reaching an initial peak apparently associated with the stage of learning letter-to-sound correspondences (3.117). There appears to be a second peak several years later, although the amount of longitudinal data we have at this point is still limited. Phonological awareness, another predictor of reading, is correlated with nonword repetition; the two differ in that phonological awareness does not predict vocabulary acquisition, whereas both are associated with subsequent reading performance (3.46).
There has been some discussion as to whether the predictive power of the nonword repetition measure comes from its memory component, or from its capacity to reflect some alternative aspect of phonological development (3.47). It seems unlikely that models of either memory or phonological processing are sufficiently developed to separate these two interpretations of the developmental data. However, the observation that interfering with the operation of the phonological loop in normal adults leads to the impaired learning of novel vocabulary, while not affecting the capacity to associate pairs of meaningful words (3.70, 3.143), is consistent with our interpretation based on the memory component of nonword repetition. This supports a single interpretation of the developmental and adult data, whereas the alternative view would appear to require different hypotheses for children and adults.
We are currently exploring the hypothesis that the phonological loop is also involved in the acquisition of syntax, using a method in which normal subjects acquire artificial grammars under conditions that facilitate or interfere with the operation of the phonological loop. Future studies will also examine the role of phonological memory in the acquisition of syntax in both normal and specific language-impaired children, in a collaborative study involving Gathercole, Bishop and Baddeley.
A2. The Visuo-Spatial Sketchpad: Apart from collaborative studies with Logie in Aberdeen (3.59, 3.135), work on this component has focused principally on the studies of conscious awareness and the vividness of imagery described below.
A3. Working Memory and Conscious Awareness: A series of more theoretically oriented papers have speculated on the role of working memory in conscious awareness (3.106, 3.111). Stimulated by the work of Teasdale on the analysis of stimulus-independent intrusive thoughts (see Teasdale's section in the Cognition and Emotion Programme), we have carried out a series of experiments in which secondary tasks were used to study factors influencing the rated vividness of visual and auditory imagery. Baddeley and Andrade began with the hypothesis that a vivid visual image was one that was represented richly within the sketchpad, while an auditory image depended crucially upon the phonological loop. A series of studies indicated, however, that a major involvement of these two subsystems occurred only when the image concerned novel material that had to be maintained in memory using the systems in question. Disruption of these subsystems had little effect on the rated vividness of images for familiar sights or sounds presumably based on information retrieved from long-term memory. The data indicated that rated vividness was a function of the amount of information potentially available to the subject, regardless of whether this limit was set by the capacity of working memory or by the amount of prior knowledge. This hypothesis was tested directly in two studies; in one, duration of exposure of an unfamiliar pattern was used to vary the amount of information in STM; the image was rated as more vivid after a long than after a short exposure. In a second study, coloured pictures of British wild birds were shown, after which subjects rated the vividness of their recalled image of the pictures; vividness correlated highly with the subjects' prior ratings of knowledge of British birds.
A4. Awareness, Cognition and Anaesthesia: After general anaesthesia patients typically have no explicit memory for surgery. There is however evidence from indirect memory tests that some learning may occur despite anaesthesia (3.8). This learning may take place when the patient is completely unconscious or during moments of consciousness which are not detected and remedied by the anaesthetist (3.8). The latter situation arises from the widespread use of neuromuscular blockers. Although these drugs aid the surgeon by paralysing the patient, they hinder the anaesthetist by abolishing the reflexes which were traditionally used to indicate when the patient was gaining consciousness.
Jones and his team in the Cambridge University Department of Anaesthesia are developing a monitor of depth of anaesthesia which will show how unconscious a patient is even if he or she has been temporarily paralysed. Essentially, this monitor records the frequency of auditory stimulation which evokes the maximal response in the patient's EEG, known as the coherent frequency. This point of maximal excitation declines as the dose of anaesthetic increases, and increases with painful stimulation. Andrade and Baddeley collaborated with Jones et al. in a study designed to validate coherent frequency against other measures of brain function (3.11, 3.68). Volunteers were anaesthetised with incremental doses of isoflurane, at concentrations low enough to permit responding to target items on two cognitive tests (word classification and within-list recognition). Assessments of coherent frequency, memory (within-list recognition), and word classification performance at each dose revealed high correlations amongst these three measures, supporting the assumption that coherent frequency reflected awareness. On recovery from the anaesthetic, subjects were given a test of recognition for some of the stimulus items presented during anaesthesia. At 0.4% isoflurane, where subjects had still been able to respond to exemplars of categories and repeated words in the working memory task, they were subsequently unable to recognise the stimuli which had been presented at that dose. Hence, amnesia for surgery should not, in itself, be taken as evidence that a patient was unconscious during surgery. This work is continuing, with particular attention to the possibility of implicit or indirect measures of memory (see below).
A5. Working Memory and Executive Control: Serial recall of a sequence of spoken digits is impaired when an irrelevant word is interposed between each digit pair, the "sandwich effect". The effect is however far less than would be expected by a simple chaining model of serial order, and reflects the operation of an attentional preprocessing stage in immediate serial recall (3.26).
In random generation, the subject is required to produce a sequence of responses, making that sequence as unpredictable as possible. We have proposed that this is a task that loads heavily on the central executive since its essence is to avoid stereotyped and routine behaviour. It has proved successful as a secondary task, leading to substantial interference with skilled behaviours like chess playing which are assumed to place heavy demands on executive processes (3.12, 3.112). Baddeley, Emslie, Kolodny and Duncan attempted to analyse random generation as part of a U.S. Air Force-funded project on executive control. We showed first of all that the typical random generation task involving verbal production of letters or numbers could be replaced by one involving key-pressing. Subjects typically have 10 keys and attempt to produce 100 random responses. Analysis suggested that the essence of the task lies in the requirement to switch retrieval plans as frequently as possible, since staying with any one plan will tend to lead to increasingly stereotyped responding. Combining key-pressing with a range of other tasks indicated:
(1) That concurrent verbal digit span interfered with key-pressing: the greater the digit load, the lower the randomness of the output.
(2) Simple counting did not interfere, whereas tasks that are known to depend on the operation of the central executive, and that are typically impaired in patients with frontal lobe damage, cause substantial interference. These included semantic category generation and performance of a fluid intelligence test.
(3) We were able to test the retrieval-plan switching hypothesis by means of an apparently simple task that has low memory load but high switching cost. This involves giving the subject a number and letter (e.g. 7-E), and requiring continuation of this series (8-F, 9-G, 10-H, 11-I, etc.). This caused virtually as much interference as a concurrent intelligence test. We hope that this may allow the development of a new clinical test of one aspect of executive function.
A6. Dual-task Performance: One assumed function of the executive is to co-ordinate simultaneous performance of separate tasks. Earlier work in collaboration with colleagues in Milan demonstrated that patients suffering from Alzheimer's disease (AD) had great difficulty co-ordinating tasks simultaneously involving the visuo-spatial sketchpad and phonological loop, even when the level of the constituent tasks was adjusted so as to make young, elderly and AD subjects perform at an equivalent level under single-task conditions. Subsequent work showed a much more marked decline in dual- rather than single-task performance as the disease progressed, an effect that did not simply reflect task difficulty (3.20, 3.116). We went on to produce a logistically more convenient paper-and-pencil version of the initial task, and to investigate the performance of patients with frontal lobe damage on this new version. Impaired performance occurs in those patients who combine frontal lobe damage with behavioural disturbance, and as such appears to reflect a different component of the frontal syndrome from those measured by the more traditional tasks of verbal fluency and the Wisconsin Card Sorting task.
B. Long-Term Memory
B1. Implicit Learning and Memory: Memory can be tested either directly (explicitly) or indirectly (implicitly). Direct tests are those such as recall or recognition, in which one deliberately tries to remember something. Indirect tests, on the other hand, are tasks in which performance will be enhanced merely by previous experience with the stimulus items. For example, subjects have a higher probability of successfully completing the fragment m _ g _ it _ d _ if they have recently been shown the word magnitude, even when they do not remember seeing that word. Indirect tests have the potential to elicit evidence for memory in cases where the subject cannot explicitly recall information (3.158), although in other cases performance on indirect tests may reflect a mixture of both unconscious (implicit) and conscious (explicit) memory (3.99).
This area has formed the topic of PhD dissertations by Green and by Kolodny. Green's thesis demonstrated methodological problems with a standard paradigm (3.51), and went on to explore the interaction between the processes underlying direct and indirect measures of retrieval. She showed that direct or explicit retrieval can block access to implicitly available information, with the result that normal subjects may be handicapped relative to amnesic patients under certain conditions (3.149).
Kolodny's doctoral research was devoted to understanding the role of conscious and unconscious processes in several learning paradigms (3.150). Investigations of unconscious processing in the Serial Reaction Time learning task indicated that the learning may involve the acquisition of simple response bias as opposed to more complex knowledge structures, as had previously been assumed. A series of experiments examined the ability of subjects to classify paintings on the basis of artistic style. Using subjects with recollective ability that is either declining (elderly adults) or negligible (anterograde amnesic patients), these studies demonstrated that intentional retrieval may be essential for such tasks. In contrast, patients and normal controls performed at comparable levels in a task involving classification of dot patterns. It appears that conscious retrieval may not be essential for simple category learning but may play a crucial role with more complex material with a greater semantic content (3.56).
B2. Recency: While the recency effect in free recall has traditionally been associated with short-term memory, the working memory model has interpreted it in terms of the use of a particular retrieval strategy that may be applied to either long- or short-term memory. This view was supported by a study in which a patient suffering from impaired STM showed an absence of recency in an immediate memory paradigm but a long-term recency effect for the recall of anagram solutions (3.89). The task of remembering where you parked your car was shown by Pinto and Baddeley (3.39) to resemble the recency effect in immediate free recall in obeying the constant ratio rule, which relates the probability of recalling an item to a combination of elapsed delay and inter-item interval. A theoretical paper by Baddeley and Hitch suggested that the recency effect could be regarded as reflecting the application of an explicit retrieval strategy that takes advantage of implicit priming effects (3.23).
B3. Memory and Schizophrenia: A collaboration to investigate memory performance in schizophrenic patients has benefited from the move of our principal collaborator, McKenna, from the Department of Psychiatry at Leeds to Fulbourn Hospital, Cambridge. Subsequent work replicated our initial finding of a high prevalence of memory deficits in schizophrenic patients (3.67), and explored in more detail the nature of the deficit, demonstrating impaired episodic and semantic memory, together with comparatively preserved phonological working memory and implicit learning (3.87, 3.137). Subsequent work involving Wilkins has further investigated the semantic memory deficit (3.35), while Baddeley and colleagues from Fulbourn have begun to approach the understanding of delusional behaviour in terms of models of autobiographical memory (3.124). The effects of schizophrenia on the various subsystems of working memory are currently being investigated in a study involving Baddeley and McKenna in Cambridge and Salamé from a new Strasbourg-based INSERM group studying schizophrenia.
C. Test Development
We have continued the process of developing clinical tests based on the cognitive psychology of memory. Investigation of semantic memory led to the SCOLP (Speed and Capacity of Language Processing) Test by Baddeley, Emslie and Nimmo-Smith (3.153). The test comprises two components, a measure of speed of sentence verification that is highly sensitive to both environmental stressors and brain damage, and a new estimate of premorbid verbal intelligence. Subjects are presented with a series of pairs, each containing one real word and one pseudoword, and are asked to "spot the word" in each pair (3.22). Since performance on this lexical decision task can be well preserved even in some types of acquired dyslexia, it seems likely to prove relatively robust to the effects of brain damage. Preliminary data based on Alzheimer's disease suggest that this is the case.
A new test of episodic memory has also been devised, the Doors and People Test (3.154). Separate subtests measure verbal recall and recognition (based on people's names), visual recall (based on copying and drawing from memory a series of shapes), and visual recognition in which the subject sees a series of coloured photographs of doors, and then has to recognise which of each set of four doors has been presented before. The test is completed by two delayed recall scores. It yields an overall episodic memory score which can be decomposed into separate visual and verbal, recall and recognition, learning and forgetting scores, all of which are standardised on the basis of a stratified population sample (3.154). The test has a wide range of applicability, from Alzheimer patients to young graduate students, and appears to be highly sensitive to the effects of normal ageing, schizophrenia and stroke. Finally, collaborative work with Dr Robin Morris of the Institute of Psychiatry has validated the sensitivity of the test to visual-verbal differences using a population of patients with left or right temporal lobectomy.
In collaboration with Susan Gathercole we have developed the nonword repetition test (described above) to a point at which it is likely to prove a valid and useful predictor of linguistic development, applicable down to the age of three (3.48). We propose to develop and standardise a new version of the test for use with all ages ranging from three-year-old children to adults.
The need for cognitive tests applicable to young children was emphasised by a recent invitation from the Partnership for Child Development to participate in a programme concerned with the possible cognitive consequences of parasitic infection in third world children (3.105). The programme was sponsored by the McDonnell Foundation, and aimed to call upon the expertise of developmental psychologists, cognitive psychologists and psychometricians. In a joint study carried out with Professor Grantham-McGregor and Dr Meeks Gardner of the University of the West Indies, we developed a series of new tests which were successfully used in a study on whipworm infection in Jamaica. Three new tests, one based on vocabulary acquisition, one on visual scanning and one on speed of sentence comprehension all proved usable under third world conditions, giving reliable data that correlated with measures of scholastic achievement (3.25).
D. Dissemination
The study of memory is one that evokes considerable intrinsic interest from the general public, and which has considerable potential for application. Dissemination to the general public has involved contributions to encyclopaedias, general scientific magazines and books of a general nature (3.104, 3.142, 3.163, 3.164, 3.167). In addition, Baddeley has attempted to link theoretical advances in the study of memory with its practical applications in the form of a text (3.1), and has completely revised Your Memory: A User's Guide, a book for the general reader (3.2).
We have also been concerned to ensure that recent developments in the psychology of memory are made available to colleagues in related disciplines (3.15, 3.108, 3.114, 3.117, 3.119, 3.143, 3.158). Perhaps the most notable recent example of our attempted dissemination is represented by a collaboration between Baddeley, Wilson and Watts to edit the Handbook of Memory Disorders (3.6). Over the last 10 to 15 years there have been great advances in our knowledge of memory disorders, together with a substantial start on the problem of how they should be assessed and treated. There is however, no single place in which the practising clinician can obtain up-to-date information on the state of the field. The handbook attempts to cover: the basic psychological and neurobiological background; an analysis of the major causes of memory deficit, both neuropsychological and neuropsychiatric; methods of assessment; and finally the current state of research on the management of memory problems. We were delighted that our colleagues in Europe and North America agreed to participate in the enterprise, and their contributions convince us that the book will be useful not only to a wide range of practitioners, but also to memory researchers.
FUTURE PROPOSALS
These are contained in a separate programme grant to be carried out after Baddeley's departure from the Applied Psychology Unit.
COMPUTATIONAL MODELLING OF SHORT-TERM MEMORY (Baddeley, Fulcher, Houghton, Norris, Page)
Introduction
Empirical studies of working memory are being complemented by a programme of computational work, designed to construct an explicit computational model of the processes in the Working Memory framework. Most of the effort has so far concentrated on modelling serial recall. Apart from their obvious implications for understanding memory and language processes, issues of serial order represent an intrinsically interesting computational problem of considerable generality.
A. Computational Modelling
In the first phase of the work, we have produced a simple connectionist model using both item-item and context-item associations in a feed-forward network. This model gives a good account of a range of basic short-term memory phenomena including primacy, recency, similarity effects and grouping. In addition to these basic phenomena, the model also simulates data on recall of lists containing a mixture of phonologically confusable and non-confusable items (Baddeley, 1968) which have proved beyond the scope of any other model in the literature (3.see Baddeley, Papagno and Norris, 3.121). Because the model simulates individual trials in a short-term memory experiment, it makes predictions about the detailed pattern of errors (transpositions, repetitions, etc.) as well as the overall serial position curve; these predictions are currently being evaluated.
Despite the architectural simplicity of the model, some aspects of its behaviour are difficult to understand. This is not simply because the model is connectionist, but because of the complexity of the interactions between its basic theoretical components. We are therefore supplementing the connectionist modelling with simple mathematical and statistical models, which help to isolate the individual components (e.g. context and item associations, item memory, phonological confusion) of the theory and to achieve a better understanding of their interactions.
The complexity of some of these interactions can be illustrated by one of the first results of this enterprise. We discovered that we could produce a classic serial position curve from a simple model incorporating the assumptions (i) that each successive list item is less strongly associated with a 'start-of-list' marker, and (ii) that there is a strong tendency for items not to be recalled twice (i.e. subjects can remember the items recalled). With noise in the recall process, this simple model generates both primacy and recency effects and gives a good account of the detailed pattern of transposition errors. Recency effects emerge partly because items that fail to be recalled in their correct position have a strong tendency to be recalled in immediately following positions, and partly because terminal items have no following items competing for recall. Recency is not dependent on any increase in strength or availability of list final items; indeed, simply increasing the strength of encoding of list-final items generally causes a decrease in recency. This mathematical model actually gives a better fit to the data than does our connectionist model and helps to identify key features which are essential properites of any effective model of serial recall.
We are currently using the knowledge gained by these simple statistical models to guide construction of further connectionist models and guide new empirical studies.
Concern with the problem of serial order is also evident in Houghton's work (Houghton, 1990). Houghton's competitive queuing model has been a major influence on other workers in the area. For example, competitive queuing formed the basis of a recent model of serial recall by Burgess and Hitch (1992).
FUTURE PROPOSALS
A. Computational Modelling
Future computational research will continue to concentrate on serial recall and chunking. Our immediate aim is to build on our success in producing a statistical/ mathematical model of serial recall and to incorporate the essential features of the mathematical model into a revised version of our connectionist model of memory.
One of our aims in this work has been to move the burden of explanation from considerations of 'architecture' to consideration of the way in which list items are represented in memory. Central to this is an investigation of the nature of the representations which support the coding of serial order. It seems impossible to account for serial recall simply by coding associations between list items (such accounts face particular problems in accounting for data from Baddeley, 1968). Associations must also be stored between items and 'context' (c.f. Schneider & Detweiler, 1987; Burgess & Hitch, 1992). However, the standard means of achieving this is to produce a contextual representation which evolves over list positions. This begs the question of how the context can be made to evolve in the same way at recall as it did at presentation. Our own work shows how serial recall could be supported largely by associations between items and context provided by the start of the list of items, thus avoiding the problems inherent in an evolving representation of context. Understanding the detailed nature of the representations supporting serial ordering is one of the fundamental issues in the explanation of a wide range of sequential behaviours and will be the major theoretical concern in future work on modelling short-term memory.
Currently our own models, as with TODAM, assume that subjects in a serial recall task are able to remember list items that they have already recalled so as to avoid repetitions. At present this is an assumption built into the model and the model does not have a memory for recalled items. We intend to add this component to the model and to examine subjects' memory for recalled items in experiments where lists include item repetitions. The behaviour of the model will be compared with a detailed analysis of the nature of errors and transpositions made by subjects.
We also intend to examine the relation between short- and long-term memory to study issues such as chunking and the role of representations in long-term memory in supporting short-term recall. For example, long-term representations may enhance short-term recall by supporting short-term memory representations, by reintegration (i.e. at retrieval), or by chunking. Given only a verbal expression, these alternative accounts are difficult to tease apart. Computational modelling can provide a clearer theoretical understanding of the implications of these different mechanisms.
In all of these cases, the computational modelling will be carried out in close association with experimental work. Deciding between alternative models often demands a fine grained analysis of error data. Serial position curves alone, which are typically all that is reported in the literature, are insufficient. This means that we have to collect our own data because, even when simulating well known phenomena, adequately detailed analyses of the data are rarely available.
CONNECTIONIST MODELS OF LONG-TERM FORGETTING AND AMNESIA (Baddeley, Murre)
Overview and cooperation
In the past two years, work was carried out on the development of several approaches to modelling long-term forgetting and amnesia. Most of these models are connectionist. One model is based on multi-dimensional scaling. Much of this work was carried out in cooperation with Baddeley. Work on long-term recovery of seemingly lost memories, such as a second language not used in many years, is undertaken in cooperation with Atkins. One of the models developed for amnesia has been generalised and is currently being applied in a more general theory of rehabilitation of brain injury, a project that is pursued jointly with Ian Robertson. For many of the technical aspects of neural network modelling, I have cooperated with Happel and Heemskerk at Leiden University.
A. Forgetting in LTM (Murre, Baddeley)
Neural networks do not generally show psychologically plausible forgetting curves. The most popular neural network paradigm, backpropagation, shows catastrophic forgetting: newly learned patterns lead to complete forgetting of existing memories. This effect was investigated and its main cause determined. One method of reducing the hidden-layer overlap is by occasional rehearsal (retraining) of earlier items APU 3185. This learning scheme is very similar to the learning method of expanding rehearsal, which is one of the most efficient learning strategies known.
Further exploration of forgetting in connectionist networks suggested that even very simple networks, without hidden layers are able to model human data. A series of simulations with a two-layer backpropagation network (only an input and an output layer) confirmed this intuition. The type of interference patterns obtained in these simulations mimic exactly those summarised in a classic paper on the role of similarity in interference in human long-term memory by Osgood (1949). The model was also successful in accounting for the effects of context-dependency of the type described by Godden and Baddeley (1975).
Another type of connectionist model was investigated that is more biologically plausible, without sacrificing psychological plausibility. These network models (Amari networks) use only positive activations and simple Hebbian learning. On the surface, learning in this model is completely different from that just described. Nevertheless, it can be shown that the basic effects of interference, described above, carry over to this class of models as well. That advantage of Amari models is that, compared to backpropagation models, both the structure and the learning method is in better agreement with current neuroanatomical models of the cerebral cortex.
An interesting aspect of our implementation of the Amari models is that a limited number of learned patterns (items) can become active at the same time. This is not possible in most other neural networks, where only a single pattern can be active at a given time. This co-activation of patterns is a plausible, and perhaps necessary, mechanism for the formation of item-item associations. The number of patterns that can be activated simultaneously is dependent on the total activation allowed. In this way, the total activation regulates something akin to a short-term memory buffer. There is not really a buffer in the networks because all patterns are distributed. The 'buffer' is a limit on the number of co-activated items that emerges through an overall limit on the activation level. When presented with a series of learned patterns under a new context, the first few items become attached more strongly to each other and to the context. This leads to the familiar primacy effect in free recall, simulated by activating the list context only. (This will then lead to recall of some of the associated items.) Since my primary aim was not to model working memory but amnesia, this aspect has not yet been further explored.
B. Amnesia
The Amari model described above was used to test several hypotheses about the temporal gradient in retrograde amnesia. Though the evidence for the preservation of old over recent memories is disputed, there seem to be several clear-cut cases where old memories were preserved better than recent memories. Simulations suggest that surviving older memories may have recruited additional nodes (neurons) and that these function as an additional 'support network'.
Rather than continuing with detailed simulations of specific effects in retrograde amnesia, we decided to leave the existing model aside for the moment and rather concentrate on the relation between anterograde amnesia (AA) and retrograde amnesia (RA), as well as on various other forms of amnesia (including transient global amnesia, TGA). The literature suggests a complicated pattern of dissociations. The few existing computational models of amnesia all assume direct coupling of AA and RA, or they ignore the existence of RA altogether. This is a very unsatisfactory state of affairs. To accommodate the more complicated pattern of dissociations that are found, a blueprint for a new model was developed: the trace-link model. In its current form this is a global model comprising a trace system (simulated 'cortex'), link system (simulated 'anterior/medial temporal cortex'), and an modulator system (simulated 'hippocampus'). A new memory is formed by associating existing patterns in the trace system. The trace system is basically the Amari network described in the last paragraph, but in this model is thought to be many times larger and structured into different modules. Initially, the link system binds together the elements of a new memory. With time, associations will form within the trace system (supporting structures), and the memory becomes independent of the link-system. The modulator system is necessary to provide the plasticity to the connections in the link system. It causes increased learning in response to significant or novel events. The trace-link model can be shown to predict most of the dissociations between AA and RA. It can also deal with TGA, shrinkage, preservation of implicit memory under AA, and with temporal memory-gradients. It also gives a possible explanation for semantic dementia.
B1. Categorisation in Amnesia (Kolodny, Murre): When modelling forgetting, it important to study the underlying memory representations in the models. In particular, it is vital to know which types of categories can still be learned by amnesic patients. Kolodny has recently shown that amnesic patients are able to learn to categorise novel categories of dot patterns, but not artistic styles. This suggest that part of the processes involved in categorisation are based on explicit memory, whereas others may be involve implicit memory.
To get a clearer view on the processes underlying categorisation, existing models for the formation of categories were studied. This work builds on earlier work carried out at Leiden University (Murre, 1992). Simulations were carried out with an extension of Nosofsky's 'Generalised Context Model' for categorisation. This model is based on the formation of psychological spaces. All items in a category are remembered. Categorisation is carried out by comparing a new item to all items in a category. We developed an extended model which performed significantly better than Nosofsky's own model (using his data and taking into account the extra parameter available in the extended model). A connectionist implementation of this model has also been investigated.
C. Memory Repair
C1. Recovery in LTM: A model of forgetting and amnesia might allow certain predictions about the recovery of partially lost memories. Such recovery occurs in everyday life, when we visit France and have to speak French again after not having used the language for, say, ten years. It may also occur when recovering from brain damage whereby information and structure has been lost. As part of this project, we investigated recovery in the backpropagation model described under a). Relearning of part of a degraded set of words showed significant improvement in the performance of the rest of the set, even though that part of the set had not been relearned. A similar effect has been reported by Hinton for a different model (the Boltzmann machine), with a different type of disturbance (noise rather than interference), and with different patterns. In cooperation with Atkins we have been systematically exploring the extent and characteristics of this phenomenon in neural networks and in human subjects.
C2. Recovery from Brain Damage: In the brain, memories are under continuous threat of degradation by volatile neurochemical processes. It is likely that the brain invokes a dynamic strategy of continuous self-repair to restore lost information as soon as possible. In this hypothesised process, extensive use is being made of the redundancy of neural representations (as dense networks). Using the Amari network described above, several such strategies were tested based on continuing recruitment of neurons and the formation of new connections. Preliminary simulations and mathematical analysis show that memories are much more robust to lesions when even primitive self-repair processes are introduced. There is a critical lesion factor after which most memories will be beyond repair. Both mathematical analysis and simulations show that self-repair processes are very effective in prolonging the life of memories.
The idea of continuous repair of memories fits in well with Robertson's theories of rehabilitation, where cognitive functions consisting of many modules must be restored. Recent advances in neuroscience show that the brain is much more plastic than generally believed and that there are, therefore, good grounds for proposing more dynamic models of rehabilitation than those based on mere compensation strategies. We are currently working on a global connectionist model of rehabilitation which is described more fully in Robertson's section.
D. Technical Aspects of Neurosimulation
The practice of computer simulations with neural networks involves at least three technical areas: (1) optimisation of the architecture and parameters of a model, (2) development of appropriate simulation software (neurosimulators), and - possibly - (3) implementation in parallel hardware for neural networks. Because it is believed by many that neurocomputers will complement existing serial and parallel computers in the near future, these technical aspects continue to attract much interest by the engineering community. The technical aspects outlined below may, therefore, find application outside the realm of modelling and theory.
D1. Optimisation of Parameters and Architecture by Genetic Algorithms (with Happel, Leiden University): Genetic algorithms are search methods based on genetic evolution. We applied them to the practical task of finding the optimal parameter set and architecture for a model for handwritten-digit recognition. Prestructuring of a neural network in terms of broad connectivity patterns can be shown to greatly increase the learning speed and final performance (3.52).
D2. Development of Software for Neural Networks: Neurosimulators: Good software for simulation is indispensable. Modellers spend a large proportion of their time developing software. Earlier, in Leiden, several simulation environments were developed. One of these, MetaNet, makes extensive use of a graphical interface. This environment is extensive (30,000 lines of C-code) and can be used for many different neural network paradigms. It is, however, not yet fully functional. To increase the functionality of MetaNet by incorporating successful elements of other systems in academia, a large survey of existing neurosimulators was undertaken. Via repeated inquiries on e-mail, references to more than 100 systems were assembled, probably the most comprehensive database on neurosimulators. A review paper on 40 of these systems will appear in the 'Handbook of Brain Research and Neural Networks' (3.139).
D3. Implementation in Parallel Hardware: Neural networks are parallel distributed systems. Implementation of these models in parallel hardware can greatly speed up simulations. This is one of the reasons why the development of a parallel neurocomputer was undertaken in Leiden University (starting in 1989). This machine, the Brain Style Processor, has 400 processors, each of which implements the CALM algorithm developed by the author. This machine has recently been completed (3.53).
Design principles of very large neurocomputers have strong parallels with the design of the brain. For example, the problem of connecting millions of 'neurons' becomes an important issue. The connectivity problem also forms the basis of the model for amnesia outlined above. Implementation in existing parallel computers (transputer networks) was analysed from the perspective of various patterns of connectivity. It was proven that modular networks show better implementation characteristics than random neural networks.
FUTURE PROPOSALS
A. Forgetting in LTM
The forgetting model will be extended, so that it can better deal with the effects of context. Once this problem has been resolved, specific simulations will be carried out of the following effects: pro-active interference in the Brown-Peterson task, release from pro-active interference (by change in semantic category), and further aspects of retroactive interference. Preliminary analyses of the current, simple model have shown that certain aspects that have led to controversy in the 1970s (e.g., item-independence) can be tackled in a mathematical fashion. This line of research will be continued, taking into account inter-subject variability. Forgetting will also be investigated in the Amari network. If both the problem of context and interference in the Amari network are resolved, serial position curves will be generated in a simulated free-recall paradigm.
B. Amnesia
The present status of the model is that it characterises several findings in amnesia in a global manner. The next stage is to develop a more detailed model. The final aim is to integrate the global model with the Amari network, in this way arriving at a model that is capable of explaining both normal forgetting and (when lesioned) pathological forgetting (amnesia). The studies described earlier will enable rapid development of such a model. Further refinements must subsequently be carried out, based on specific patient data. A specific topic to be explored is preserved implicit memory under retrograde amnesia. Because few patient data are available, the model might be used to yield specific predictions which can then be tested on patients.
B1. Categorisation in Amnesia: Integration of the categorisation model with the amnesia model will be undertaken. The amnesia model has been designed with this purpose in mind, among others. A specific task is to model Kolodny's data on categorisation. This exercise will shed more light on the question of what processes (implicit or explicit memory) are involved in categorisation of probabilistic stimuli. Further psychological experiments with amnesic patients will be necessary to be able to develop a fully articulated theory on this issue.
C. Memory Repair
C1. Recovery in LTM: Recovery of seemingly lost memories will be pursued further in cooperation with Atkins. Several experiments will be carried out to establish under what conditions the phenomenon occurs. In addition, single-case studies will be carried out with extensive - but controlled -exposure to a 'forgotten' language, such as Chinese learned over ten years ago and not used since then. The hypothesis is that partial exposure will result in improved performance on the rest of the language, to which the subject has not yet been exposed. This plausible hypothesis still remains to be validated under controlled circumstances. If demonstrated, the corresponding effect in neural networks with hidden representations will be explored in backpropagation and Boltzmann machine models.
D. Technical Aspects of Neurosimulation
D1. Optimisation of Parameters and Architecture by Genetic Algorithms: This line of research will not be pursued further. The software and methods developed so far are adequate to act as tools in further modelling research.
D2. Development of Software for Neural Networks: Neurosimulators: MetaNet will be developed into a fully functional system. Several partial systems will be constructed first, to test the feasibility of the solutions. The final aim will be to arrive at a flexible system that is first of all of use to modellers in psychology.
D3. Implementation in Parallel Hardware: We will continue to study the connectivity problem in neurocomputers and in the brain, in cooperation with Heemskerk in Leiden. At present, a design for a very-large neurocomputer based on a fractal architecture is being developed. This work builds directly on previous work on the Brain Style Processor, which has a modular architecture. We will compare the hypothetical performance of a variety of neural network architectures on a range of processor topologies (including fractal, modular, and grid) using both analysis and simulation.
CAUSALITY, CLASSIFICATION AND IMPLICIT LEARNING (Shanks)
A. Judgement of Causality
Shanks has been exploring the processes that control the perception and judgement of causality in humans (3.73, 3.76, 3.78, 3.81, 3.82, 3.131). Studies have shown that perceived causality in a Michottean collision task can be dissociated from cognitive judgements of causality (3.73). It has been argued, on the basis of correspondences between conditioning and causal judgement (3.76, 3.81, 3.131), that conditioning is a mechanism that has evolved in order to pick out causal relationships in the world.
A debate has taken place on the extent to which causality judgements are normatively accurate (3.82), and experiments have been conducted (3.82) which question the assumption of normativity. For instance, judgements are strongly affected by the order in which trials are presented.
B. Classification
Shanks has carried out a number of empirical studies of human category learning (3.145, 3.77, 3.79, 3.80, 3.174) aimed at testing connectionist, rule-based, and exemplar models. These tests have successfully discriminated between the three classes of theories, with connectionist models being generally supported.
Theoretical work on the topic of category learning (3.151) started from the observation that models of learning based on the backpropagation learning algorithm and architecture suffer from two major flaws: (i) they cannot account for the effects of selective attention on learning, and (ii) they suffer from catastrophic forgetting. An example of the former is the finding that it is easier to classify stimuli if only one dimension is relevant to the classification than if two dimensions are relevant. Evidence of catastrophic interference comes from demonstrations that information is completely eradicated in backpropagation networks when later, partially-overlapping information is encoded. Although humans do of course show forgetting in such circumstances, they are affected to nothing like this catastrophic degree.
As an alternative to backpropagation, Shanks (3.151) developed a new model called the consequential region model which avoids these problems and generates excellent fits to published category learning data. In the model, stimuli are represented by the activation of large numbers of hidden units whose receptive fields correspond to hypotheses concerning the size and shape of the category in the input space. A process of selection, via the delta rule, picks out those hidden units which are most predictive of category membership. The results illustrate how an associative network can show appropriate sensitivity to inter-item similarities among training exemplars as an emergent property of its scheme for representing stimuli.
C. Implicit Learning
Another topic that has been of major interest to students of associative learning is the question of whether learning can occur implicitly. The notion that there exist independent explicit and implicit learning systems combines two further distinctions, (i) between learning that takes place with, versus without, concurrent awareness, and (ii) between learning that involves the encoding of instances (or fragments) versus the induction of abstract rules or hypotheses. Implicit learning is assumed to involve unconscious rule learning (3.173).
Experimental work has been undertaken (3.84, 3.51, 3.85) which questions some published claims, from reaction time and system control studies, concerning implicit learning. For instance, we (3.51) were unable to replicate some of Broadbent's evidence for implicit learning.
In an extensive review, Shanks (3.152, 3.85) examined the evidence for implicit learning from studies of subliminal perception, conditioning, artificial grammar learning, instrumental learning, and reaction times in sequence learning, and concluded that unconscious learning has not been satisfactorily established in any of these areas. The assumption that learning in some of these tasks (e.g., artificial grammar learning) is predominantly based on rule abstraction is also questionable. When subjects cannot report the "implicitly learned" rules that govern stimulus selection, this is often because their knowledge consists of instances of fragments of the training stimuli rather than rules. In contrast to the distinction between conscious and unconscious learning, the distinction between instance and rule learning is a sound and meaningful way of taxonomizing human learning. Various computational models of these two forms of learning have been considered.
REMEMBERING SPECIFIC EVENTS (Bekerian)
Introduction: Long-Term Memory for Specific Events
Memory represents one of the most essential cognitive abilities for normal human functioning. We constantly rely on our ability to refer to the past and to use previously acquired information. Our research has attempted to provide some insight into this ability. The research is unified by a concern for naturalistic situations where remembering -- what is claimed to have happened in the past -- determines the fate of lives. In the most dramatic case, for example, we examine how people remember physical and sexual abuse in evidential interviews.
The research relies on data from laboratory experiments and observational studies. Such converging methodologies are not only essential from the standpoint of application of the research. Importantly, the use of different methodologies has allowed us to maintain the integrity of theoretical arguments when they are applicable, and extend and/or qualify such arguments in the face of specific naturalistic contexts.
The application of the research has been particularly successful, especially with regard to issues of public health. These will be discussed more fully below. We merely note here that thus far the application of the research has ranged from the development of mnemonic aids through advancements in information technology, the training and assessment of professionals responsible for the interviewing of alleged victims of child abuse, to the effectiveness and dangers of therapeutic intervention for victims of violent crimes.
The research retains notable objectivity in discussing otherwise highly sensitive and politically dynamic issues, for example assessments of Home Office guidelines for the video interviewing of child witnesses. This has resulted in the Bekerian being approached to overview highly productive, and highly controversial areas of applied memory research, for example eyewitness identification, and offender profiling. Apart from the main research interests, Bekerian also serves as an advisor on cases of appeal (for example, JUSTICE) as well as general Police enquiries. Because of the obvious applicability and importance of our research to issues of public concern, we have also been involved in the recording and production of BBC educational television, and national television and radio programmes. In each case, Bekerian provided expert opinion on applied memory issues.
The general themes that form the theoretical basis for the research will be discussed under three major topic headings: a) the malleability of memory b) individual variability in memory, and c) the effects of remembering environment.
A. The Malleability of Memory (Bekerian, Shaw, McCubbin)
Remembering changes over time: old details drop out; new details are included. The stability, and instability of memory is of great theoretical debate. For example, stability in the information recalled is assumed to indicate when an account is based on a real event, as opposed to confabulated memory. Such considerations also form the basis for the application of cognitive theories of memory in many criminal cases. For example, any inconsistencies in a witness' account may cause grounds for disbelief, unless these changes can be assumed to be part of "normal" remembering. Knowing when, why and how, remembering changes are important.
Our research has focused on why memory improves over time. We have looked specifically at the phenomenon of hypermnesia, where a person remembers more new information than s/he forgets (3.86). The research has involved a series of experiments that have used interference paradigms, looking at the effects of interpolated activity on memory for concrete word pairs. The work is the first to use the interference paradigm to investigate hypermnesia, so the results are of particular importance. One of the major explanations for hypermnesia is that imagery at the time of learning is responsible for hypermnesia. Our experiments have shown that although imagery at encoding is a necessary condition for hypermnesia (abstract words do not show memory improvement), it is not a sufficient one. Rather, the results suggest that the distinctiveness of the information seem more important. For example, when interpolated material is processed in a manner similar to that of the original material, no hypermnesia occurs, regardless of whether the interpolated materials are processed at "deep" levels (interactive imagery) or "shallow" (separate imagery) levels.
Other work in progress has extended our interest in hypermnesia and the interference paradigm, looking at the effects of emotional arousal on memory improvement (3.101). There is some considerable controversy over the effects of arousal on memory. Some argue that arousal interferes with processing capacity, thereby reducing the "strength" and type of information originally encoded; others argue that arousal can enhance the processing and subsequent retention of information over time. Our findings suggest that negative arousal interrupts hypermnesia, but only when it occurs during interpolated trials, not when it occurs prior to original learning, suggesting, as before, the importance of the interpolated activity. These findings are important for discussions of inconsistencies in real victim's accounts of trauma. For example, if it is the case that hypermnesia can occur for events initially experienced with great trauma, certain inconsistencies in a victim's account may not be grounds for great concern.
B. Individual Variability in Memory (Bekerian, Barnard, Eldridge, MacLeod, Williams)
Research in this area has highlighted the importance of individual variability in the way that information is remembered, and in the manner in which information might be represented (3.42, 3.60). One argument is that individuals differ significantly in terms of the level and content of their representation of their daily lives. A novel deviation of the research is that it has focused on time periods of life that have received little theoretical attention (3.42). For example, the research examines schemas for a person's working day, using both free and cued recall paradigms. Schemas are generally discussed in terms of very global structures, such as life-time periods, or single events, like going to the dentist's. In contrast, the schema for a "working day" is at a more intermediate level: many single events make up a "day". The findings suggest that the degree of elaboration and content of an individual's schema can predict memory performance, and forgetting (3.42). For example, people with more elaborate schemas retain more information about typical activities than people with less elaborated schemas. These represent relatively novel discussions of autobiographical memory phenomena, which typically do not consider individual differences in this manner.
Current work is examining the fluctuations in an individual's schemas and the degree of elaboration of such representations (Bekerian, Eldridge, Barnard). The hypothesis is that everyday schemas undergo constant updating and modification. Consequently, the pattern of memory, even for daily events, may fail to be consistent over time. Importantly, similar arguments concerning the unstable nature of representations have been made in the context of semantic memory research; but, not within episodic memory research.
The practical applications of this work have been through the MRC's involvement with Rank Xerox EUROPARC, on the development of automatic mnemonic aids for the work environment. As the practical demands placed upon memory in the workplace increase, the potential uses of information technology to support memory become more important. The MRC have been responsible for providing normative information about what people remember of their working life illustrating in the first instance the nature of memory problems that people might have. For example, we have shown that people's memory for institutional talks (such as seminars, or formal expert talks) is appalling after only a week (sometimes at chance), and that the reinstatement of cues, like overhead transparencies used by the speaker, fail to improve performance. These data suggest that the potential application of automatic mnemonic aids may have to be restricted to "simpler" memory problems, such as object location (e.g., finding the folder in which you saved a particular electronic file).
C. Effects of the Remembering Environment (Bekerian, Dennett, Dritschel, Toplis, Wing)
C1. Empirical Studies: Over time, people may have to remember the same event across different circumstances, such as recounting a first date to one's mother and, then, to one's best friend. These circumstances, which we term the 'remembering environment", have been of particular interest, in that the remembering environment is, potentially, manipulable in real world situations. One of our concerns has been the predominance in experimental investigations of what has come to be called episodic remembering (3.130): the person is requested to recount a specific event that occurred in the past, and is intentionally engaging in the act of remembering. This type of remembering parallels important characteristics of many real world situations; for example, in an evidential interview, the interviewer requests the witness to engage in episodic remembering. However, there are other conditions under which people remember events. For example spontaneous remindings that occur in the course of a conversation seem more like the result of some "priming" mechanism, rather than any intention to remember. Similarly, phenomena such as flashbacks, where there is an unrequested, unintentional and unwanted remembering of an event are also outside the scope of most theories of autobiographical memory.
The research has identified the effects of subtle differences in the types of remembering environments. For example, a simple manipulation was introduced whereby the person was asked either to speak, or write, the account (3.32). The findings showed that speakers produced more correct information than writers. We are currently pursuing the possibility that writers suffer interference of Working Memory processes, due to the demands placed on them by the writing task itself. For example, writers may be at a disadvantage, since the visuo-spatial scratchpad is otherwise occupied by the task of writing. Imagery instructions, which encourage the formation of a visual image prior to output, would be expected to have an effect on writers, but perhaps not speakers, if this is correct. Preliminary analyses have revealed that imagery instructions do effect speakers, both in terms of correct as well as incorrect information that is recalled (3.129). We have already considered the serious implications these results have for real interview techniques that rely on imagery as a technique to reinstate context (3.33). For example, encouragement to image may result in the person falsely "remembering", with confidence, details that were never present.
We have also been concerned with the theoretical debate over whether accounts based on real memories differ quantitatively and qualitatively from those based on fictionalised memories. The practical implications of this debate are immense. If it were true that certain features in an account can distinguish real from confabulated memories, the application would be seen across the legal/civil domain, ranging from witness reports to accusations of 'recovered' memories of sexual abuse.
Our review of the evidence has suggested that certain criteria seem to be uniformly regarded as indices of "real" memories. These criteria include the presence of perceptual detail, the spontaneous nature of the account, and an unstructured narrative style. We have argued that the remembering environment will influence the presence of these criteria independently of the "truthfulness" of the account. For example, we have reasoned that frequently experienced events are likely to be reported first in a schematised fashion, lacking extensive perceptual detail. Unless directed otherwise, the person may fail to provide a specific episode; alternatively, the person may be unable to recount a specific episode without some prompting, e.g., tell me about an event, x, where something unusual or surprising happened. Our work in this area has been extended directly to observations of real allegations of abuse, which will be discussed below.
Other work has examined experimentally what determines whether a person remembers a specific incident in response to a question, or whether they remember some "schematic" memory (3.95). We have used a dual task paradigm, with one task being question-answering, and the other, tapping to a rhythmic beat. The tapping task requires some executive capacity, so puts a cognitive load on the person when performed with another task (relative to the single task conditions). The questions were everyday, common routines, increasing in novelty (i.e., deviation in the script). For example, a common routine question is "How do you brush your teeth with a toothbrush?" The least deviant question is "How do you brush you teeth if you have to use your finger?" The most deviant question is "How do you brush your teeth if you only have a stone?" We have shown that schematic memories are more likely to be remembered when the person is engaged in two tasks (i.e., when the cognitive "load" is greater), for common and least deviant questions. This suggests that it is easier to remember schematic memories particularly when under some stress. Further, we have shown that specific autobiographical memories are infrequently remembered, and this seems to be related to whether they match the question fairly precisely (e.g., I once had to brush my teeth with my finger), or, alternatively, whether they have been recently experienced (e.g., experienced shortly before the testing session). These findings provide empirical support for our argument about the effects of frequently experienced events and the schematic nature of memory. In addition they provide new data on the quality of autobiographical memory under cognitively demanding conditions.
C2. Naturalistic Observations: The theoretical arguments emerging from the empirical work have been developed in parallel with detailed analyses of therapeutic interviews and evidential interviews with real victims of serious crimes (3.125, 3.126, 3.127, 3.128, 3.146, 3.147, 3.168, 3.169), in the main, alleged victims of child abuse. There has been extensive collaboration between Bekerian/Dennett and the Statutory and Caring Agencies responsible for the interviewing of such victims. These Agencies have provided Bekerian/Dennett with video-tapes, transcripts and statements of real interviews, as well as case-relevant details.
The implications of this research for the assessment of interviewing skills should not be underestimated. Ineffective interviewing of victims of trauma or serious crimes is an issue of general public and social concern. It is costly, in terms of man-hours, in terms of expenditure of otherwise limited training budgets, and in terms of human discomfort. For example, of all cases of child abuse passed to the Crown Prosecution Service (CPS) by the police (approximately 250 in the region a year), only seven percent are considered for prosecution (about 18 cases). Of those 18 cases, only one is likely to sustain a successful prosecution. The CPS cite poor interviewing techniques as one of the primary reasons for this surprising discrepancy.
Procedures for analysing both the interviewer's and the victim's behaviour have been an essential component of this work. The analyses address both practical and theoretical issues. For example, we have devised a means of depicting interviewer behaviour in a way that can distinguish evidential style interviewing from therapeutic style interviewing (3.146, 3.147, 3.128). This distinction is critical in practice. Home Office guidelines stipulate against the use of therapeutic interviewing for alleged child abuse victims. Theoretically, the distinction between styles of interviewing is important, as the distinction forms the basis for arguments in the "repressed memory syndrome" debate. For example, therapeutic style interviews, because of their emphasis on emotional interpretation, might be expected to yield highly reconstructed and embellished accounts, rather than highly veridical ones. Because the data-base is derived from actual interviews, the work has the advantage of directly linking theory with practice.
We have focused on factors that are likely to influence episodic remembering of child abuse (3.125). In particular, we have examined the evidence regarding the effectiveness of different interview techniques, particularly those which claim to enhance memory without simultaneously increasing incorrect recall (3.33, 3.126). For example, we have argued that mnemonic techniques should be introduced later, rather than earlier in repeated interviews, simply because such techniques are likely to introduce errors into the account as well as enhancing correct recall (3.33). Additionally, we have identified potential advantages and disadvantages of using mnemonic techniques with specific subject populations and event scenarios (3.33, 3.125, 3.126, 3.127, 3.129). For example, encouraging the use of imagery with a suggestible witness (e.g., a young child of four) is considered a dangerous tactic, given it may promote incorrect details being recalled.
The development of procedures for analysing interviewer/victim behaviour is a critical achievement (see 3.42, 3.128). No other procedures for analysing memory in evidential interviews are available within the published literature, rendering the work unique in its status (3.128). Although these were developed in the context of our collaborative work on child abuse, the application to other dyadic conversations is straightforward, and serves as the grounds for future research into therapeutic intervention.
Our procedures and analyses of interviews have already been incorporated into interviewer training programmes on violent/traumatic crimes that are conducted by the County Statutory and Caring Agencies (3.146, 3.147). Reactions from these Agencies have been unreservedly favourable, as the procedures have proven successful in identifying areas of strengths and weaknesses in an interviewer's skill. The work is now being considered by Regional Health Authority, and this future work will be discussed
The issue of validation has been particularly important, for obvious reasons. Thus far, we have considered the effectiveness of our input to training in the following way. There are certain types of interviewing that both cognitive theory, and the law, say an interviewer shouldn't use: these include leading questions, closed questions, questions based on emotional interpretation, coercive questions, sudden changes in topic, and so on. People who interview in that way are considered "poor". Part of our assessment has been to identify these more "undesirable" styles, and compare the styles of interviewers who either have or have not been on formal courses, in order to determine whether formal training has any effects.
The success of the work has led to extensive involvement with and membership on national and Home Office working parties that are responsible for the review and recommendation of working practices/guidelines for the interviewing of victims of serious crimes (Association of Chief Police Officers, Personnel and Training Committee, Working Party on Investigative Interviewing; Association of Chief Police Officers, Steering Committee on Offender Profiling; Home Office Video Guidelines for the Interviewing of Child Witnesses). Additionally, Bekerian is a member of the BPS Working Party for the Recovered Memory Syndrome, which will provide a report regarding the nature of the false memory syndrome in September, 1994. Most recently, Bekerian has been asked to provide a peer commentary over the debates of recovered/false memories.
We have established collaborative links with Dr. J. Jackson and the Netherlands Institute for the Study of Criminality and Law Enforcement (NISCALE) concerning interview techniques of witnesses/suspects. Prof. U. Undeutsch (University of Koln, Koln, Germany) has also been an important collaborator in research concerning the assessments of truth in statements of alleged child abuse victims (see 3.125, 3.127), providing us with statements of evidential interviews with alleged victims.
Work in progress is involved in the systematic review of current practices in the investigation of serial crimes, focusing on interview techniques that are used in the United States, Great Britain, the Continent and Australia.
FUTURE PROPOSALS
A. The Malleability of Memory
Changes in "Self": People have a notion of a "self" which represents their own personal set of beliefs, values and attitudes. Autobiographical memory largely relies on this notion, as without a "self" the concept of autobiographical memory becomes somewhat curious (3.130; see Barnard & Teasdale, 1991; Bermùdez, Marcel & Eilan, in press). Our particular concern here is how changes in the "self" effect what is remembered of a highly traumatic event. Specifically, we define changes in "self" as remission from Post-Traumatic Stress Disorder (PTSD).
The project uses clinical populations who are victims of rape suffering from Post-Traumatic Stress Disorder (PTSD) and are referred for therapeutic treatment. Therapeutic sessions require the client to remember the event as part of the therapeutic intervention technique. Changes in emotional reactions are monitored through the use of anxiety ratings. Discourse analyses of clients' memories will be based on procedures that have already been proven successful in similar contexts (see 3.128). Analyses will identify quantitative changes in memory, and qualitative changes in memory, e.g., changes in metaphor, personalised nature of the information.
Control (i.e., non-clinical) populations will be asked to remember significant life events across repeated sessions. Personal ratings of the significance, emotional valence of and emotional reaction to the event will be taken. These data will provide some indication of the spontaneous changes that occur in remembering. Given that the therapeutic intervention is very non-directive, the comparison with control data is somewhat less problematic.
The findings will be of importance to theories of self and autobiographical memory, illustrating the extent and nature of the influence that changes in self have on how people remember events from their past. Additionally, the findings will be useful in informing us about the "normal" changes that might be expected to occur in memory of trauma, extending our interest in phenomena such as hypermnesia.
The work also has implications for more general discussions of the effects of therapeutic intervention on memory, for example, the "false memory syndrome", where therapeutic intervention is accused of implanting false memories of abuse. For example, recent work suggests that memories based on real events are largely invariant, with only "false" memories showing the sudden recovery of details. Given the data collected on PTSD victims will be based on real events, any changes in remembering will be particularly informative towards this debate.
The work will be done in collaboration with Prof. E. Foa (Rape Crisis Center, Philadelphia), who will provide necessary transcripts of therapeutic interviews. (It is likely that T. Dalgleish (MRC APU) will be a potential collaborator, given his expertise in PTSD).
B. Individual Variability in Memory
The role of language in autobiographical remembering. Our previous discussions of childhood remembering (3.125, 3.127, 3.168, 3.169) have suggested that verbal fluency will determine whether certain mnemonic techniques are appropriate. This proposed work intends to examine the relationship between normal language development, and the emergence of autobiographical memory. There is currently a question whether fluency in language is a necessary condition for the normal development of autobiographical memory in children. The proposed work intends to examine this possibility through investigations with language impaired individuals. In particular, it is hypothesised that language impaired individuals should show marked deficits in quantitative and qualitative characteristics of autobiographical memory that are normal for the age group. Initially, the work will adopt a cueing paradigm to look at autobiographical remembering, in that this paradigm has been widely used with both adult and child populations. The work will be conducted in collaboration with Dorothy Bishop, who has access to language impaired and unimpaired children.
C. Effects of the Remembering Environment
C1 The Task of Autobiographical Remembering: What is remembered will depend upon how one queries memory, i.e., the remembering environment. Such concerns are particularly important within applied contexts, since it is critical to determine which conditions are likely to produce changes, and whether the changes are in a favourable or unfavourable direction. The project explores the possibility that as the task of remembering changes (for example, the reasons for remembering), the content of the account and its structure will vary accordingly. The suggestion that memory can vary so dramatically, depending on the task, is one which has great potential implication, both for theory as well as in the applied setting. For example, if certain remembering tasks can be found to improve memory performance, such manipulations would be easy to administer in the applied situation.
Methodologies of repeated observations and variations in the narrative demands of the task (e.g., story-telling vs. descriptive accounts) will serve as the basis for empirical work, with naturalistic or observational studies supporting the experimental paradigms. It is expected that different narrative demands will yield different memories, both in content (for example, more interpretative information) and format (for example, more personalised utterances, such as the use of "I").
Another application of these findings is with regards to the differences between the therapeutic context and other interview contexts, a concern which has already formed the basis for much of our earlier work. Therapeutic interviewing is often considered to be different from other types of interviewing because of its explicit emphasis on personal interpretation. However, an equally obvious characteristic of the therapeutic interview is that it focuses on a narrative styles of remembering, much more so than other interviews . Therefore, any effects of narrative instructions that we find under laboratory conditions will identify the consequences of adopting different remembering tasks, per se, regardless of the particular explicit nature of the interview styles.
C2. Disadvantages of Mnemonic Techniques: Remembering typically includes information that is wrong, e.g., a yellow hat is remembered as being red. The person "remembers" incorrect information, and misleads him/herself into believing that these errors are "true". The proposed research extends theoretical arguments set forward in an earlier review (3.33) and examines the consequences of the use of one particular memory enhancing technique on spontaneously generated errors. This technique, the Cognitive Interview technique (CI), is widely used in the context of evidential interviews (i.e., with victims/witnesses/suspects of crimes), therapeutic interviews, accident investigations, and in public health care studies (e.g., children's dietary habits).
Two general questions will be addressed, across a series of experiments: one, the consequences of CI with respect to qualitative characteristics of memory; second, changes occurring in memory over time. Assessments of qualitative characteristics of memory will be investigated through the use of memory attributions, or states of awareness, for example, whether a person distinctly remembers encountering the information or whether they merely know they encountered it. Generally, "remember" states are associated with higher confidence, and, in turn, are more likely to be believed as "true", when compared with "know" states. Mature and young adults, as well as adolescent and child populations will be used, since there is little evidence concerning the developmental changes that occur in the ability to make memory attributions. Delays in recall will determine whether extended intervals render discriminations more difficult. This work will be in collaboration with Prof. B. Clifford and R. Toplis (Psychology Dept., University of East London) who will provide adolescent and child populations.
It is hypothesised that CI -- relative to "standard" interview procedures -- will result in the person being biased to "remember" incorrect information and being highly confidence in this information. These effect should be more pronounced with longer delays between recall attempts. These predictions are based on assumptions about how people decide they are "remembering". For example, mental context reinstatement -- the most powerful mnemonic used by CI -- requires that the individual form a mental image of the physical environment prior to recall. With encouragement to form highly detailed contexts, the person is likely to attribute an error as being "remembered" by virtue of the extensive context under which it was produced.
Given that CI is used so extensively, any new data will have certain practical implications. For example, should CI promote more confident errors, this would have severe consequences for the use of the technique in real-world investigations. It is well established that more confident witnesses are more credible. As such, more confident errors are likely to be believed as true. This work will be in collaboration with Prof. B. Clifford and R. Toplis (Psychology Dept., University of East London) who will provide adolescent and child populations.
Most recently, the National Opinion Research Group Ltd. has solicited Bekerian's expert opinion on the use of CI for their training of market research interviewers. It is anticipated that Bekerian will be an advisor for NOP, helping them to develop training programmes, and identifying the applications of cognitive techniques like CI for market research.
C3. Naturalistic Observations of Memory: The proposed work continues the analyses of real interviews of victims of serious crimes, under both evidential and therapeutic settings (see 3.146, 3.147). Procedures used for analyses have already been developed and have provided essential information to the Statutory and Caring Agencies (i.e., County Police, Social Services, NSPCC) regarding interview skills (see APU 3.128). This work will be critical in the appraisal of current interviewing skills evaluated through specialised courses currently on offer by these Agencies, and will also determine the successfulness of training procedures in promoting better practice.. As a direct consequence of the work that has already been completed for the Agencies, the North West Anglia Health Authority is encouraging Bekerian to submit a proposal which would provide analyses of evidential interviews with alleged child abuse victims across the region. In particular, the Health Authority is concerned with the quality of interviewing currently being achieved by its professionals; and, the standard of good practice is of grave concern. The Health Authority is keen to have the involvement of Bekerian as an advisor on applied memory issues and interview techniques as well as determining the successfulness of training procedures in promoting better practice.
The County Crown Prosecution Service (CPS) has also expressed their intention to become directly involved in the project, primarily to provide input regarding the standard of interviewing as seen by legal requirements. As already mentioned the CPS are concerned with the poor quality of interviewing techniques used in child abuse cases. The CPS feel that our project can provide important information about the extent to which poor interviewing is responsible for failures to proceed with or sustain prosecution.
An evaluation of child abuse interviews is also being conducted in Leicestershire by Prof. G. Davies and his colleagues (Psychology Dept. University of Leicester), although the work has not yet begun. Prof. Davies has expressed interest in the procedures that we have developed for analyses of evidential interviews and further collaborative work is under discussion.
MEMORY AND AGEING (Maylor)
Introduction
Since my arrival at the APU in September 1992, I have continued research begun at the Age and Cognitive Performance Research Centre in Manchester. This work is primarily concerned with attempts to characterise differences in normal ageing: (i) between tasks, (ii) within tasks, and (iii) between individuals. Of particular interest is the concept of "limited impact": it may be possible to identify aspects of performance and/or individuals within a population apparently unaffected by ageing. Some of the methods used have been successfully applied in previous work on the effects of moderate doses of alcohol in healthy young adults (e.g., Rabbitt & Maylor, 1991; Maylor et al., 1992; Maylor & Rabbitt, 1993). A central issue in both areas of research is the extent to which a single factor such as reduced processing resources can account for changes in performance. For example, in speeded tasks, the question is whether the effects of old age are generally proportionate: that is, these effects may be larger in absolute terms for tasks, stimuli, or people that take longer, but are not necessarily larger in relative terms.
A. Retrieval from Long-Term Memory
It is generally assumed that normal ageing is associated with markedly-decreasing fluid intelligence (the efficiency of processing at the time of evaluation), but relatively-stable crystallised intelligence (the accumulated products of prior processing). However, it is not always the case that measures of vocabulary or general knowledge (as tests of crystallised intelligence) remain stable with increasing age. For example, there is significant age-related decline in the ability to identify a word from its definition when retrieval time is limited (Maylor, 1990b), despite age-matching on the ability to define words. It appears, therefore, that the elderly are more impaired in accessing the lexical system from the semantic system than the reverse. This is consistent with complaints by the elderly of increasing numbers of "tip-of-the-tongue" states, particularly associated with self-reported difficulty in retrieving people's names (e.g., Maylor, 1993; 3.71).
In laboratory studies of the identification of famous people from photographs (Maylor, 1990c), and television programmes from theme tunes (Maylor, 1991), the elderly are significantly impaired at recognition, naming and retrieval of semantic information. Moreover, the effect of age is more dramatic at later stages of the identification process than at earlier stages (naming and recognition, respectively). Older adults are slower than younger adults in making structural, recognition, semantic and naming decisions about faces, with the age difference in latency varying across tasks (larger for slower tasks). To examine whether this effect of ageing is simply proportionate (i.e., older adults are generally slower than younger adults by a constant proportion), the data are redrawn as a Brinley plot, in which response latencies from older subjects are plotted against the corresponding latencies from younger subjects. The resulting function is linear with a slope greater than 1, and this true for both between- and within-task variability (Maylor & Valentine, 1992).
In contrast to this pattern of proportionate slowing for choice decisions about faces, the pattern for name retrieval is disproportionate (positively accelerated) for more difficult items. This is attributable to an increased probability of retrieval blocks with age (Maylor & Valentine, 1992) that occurs for both proper and common names
The question of individual differences in ageing was addressed by studying older and younger volunteers selected as "experts" in speeded retrieval of information from long-term memory (3.63). In contrast to the earlier findings from more representative samples of the population, tip-of-the-tongue states (while retrieving either specialised or general knowledge) were no more likely to occur in the older experts than in the younger experts, suggesting that age-related decline is not inevitable but may be related to initial level of performance (see also 3.66).
Finally, two studies of long-term memory retrieval demonstrate specific age-related loss of contextual information. The first was a collaborative study conducted with Gillian Cohen (Open University) and Martin Conway (Bristol University). A year after the event, very few elderly subjects reported "flashbulb" memories for the circumstances in which they learned of Margaret Thatcher's resignation, in comparison with younger subjects (3.38). The second study examined long-term memory for auditory stimuli; there was age-related decline in recognition accompanied by recollective experience of the original learning context, but no age effect in the absence of recollective experience (3.65).
B. Prospective Memory
The ability to perform successfully in prospective memory tasks, such as remembering to pay an electricity bill on time, or to take medicine every four hours, is obviously essential for independent living. Until recently, however, prospective memory has been a somewhat neglected area of research within cognitive psychology. There are probably both practical and theoretical reasons for this, including the problems of distinguishing between memory failures and lack of compliance, between prospective and retrospective components of a task, and so on.
With regard to age and prospective memory, the elderly would not be expected to perform as well as the young because of their reduced processing resources. In particular, deficits in processes such as self-initiated retrieval, reality monitoring and output monitoring would be detrimental to performance in prospective memory tasks. However, experience and feedback accumulate with age, providing the opportunity for the elderly to develop compensatory strategies to overcome their cognitive deficits. This could account (at least partly) for the striking absence of age deficits in (1) "naturalistic" prospective memory studies conducted outside the laboratory (Maylor, 1990a), and (2) self-rated ability in everyday prospective memory tasks (Maylor, 1993). Consistent with their declining cognitive abilities, the elderly are less able to perform as well as the young under laboratory conditions, where the use of external memory aids is prevented (3.62).
From componential analyses of repeated prospective memory tasks (3.136), it is argued that performance on a particular occasion is influenced by different factors depending on whether the task is being performed for the first time (in which case, behaviour at encoding may predominate) or has already been performed (in which case, memory for the earlier occasion and its context may play an additional role). For example, the type of cue adopted in the task performed outside the laboratory (Maylor, 1990a) was found to have a stronger effect on initial performance (i.e., success or failure on the first occasion) than on subsequent forgetting (i.e., success followed by failure). On the other hand, age was found to have a stronger effect on subsequent forgetting than on initial performance. The relationship between age and forgetting in the laboratory prospective memory task (3.62) is particularly striking for two reasons: (1) it remained significant even after age differences in other measures of performance (including speed) were taken into account, and (2) it contrasts with the absence of an effect of age on forgetting in retrospective memory tasks (e.g., 3.62). Again, these findings illustrate differences in the effects of age both between and within tasks that are difficult to explain in terms of single-factor models.
FUTURE PROPOSALS
A. Retrieval from Long-Term Memory
Preliminary attempts to investigate factors (including age) associated with variability in long-term memory retrieval from one occasion to another were disappointing because of too few instances of successful retrieval followed by failure. Plans to resolve this issue include a considerable increase in the time interval between testing occasions (which has the additional advantage of minimising the influence of episodic memory).
B. Prospective Memory
Thus far, there has been little work on either the reliability or the validity of prospective memory measures. On the question of reliability, it seems plausible, for example, that prospective memory paradigms with multiple observations would be preferable to those with single observations. However, this may be a mistaken assumption, since performance on any given trial in a prospective memory task may not necessarily be independent of performance on previous trials (3.136). Also, some evidence suggests that single observations may be more reliable in the young than in the elderly, the latter group displaying greater variability from trial to trial (3.62). Future plans therefore include examining the relationships between performance in single- and multiple-observation prospective memory paradigms, and between performance in similar tasks on different occasions, both as a function of age.
On the question of validity, one of the more promising laboratory procedures is that of embedding a prospective memory task (e.g., "Remember to note the people wearing glasses") within a demanding retrospective memory task (e.g., "Name the famous faces") (see 3.62, 3.136). It is particularly striking that the reasons given by subjects for failing the prospective memory element correspond well with those given by subjects failing prospective memory tasks outside the laboratory (e.g., "I was absorbed in another task"). This suggests that the former may be a good analogue of the latter; however, it is clearly important to establish more formally that current laboratory paradigms are indeed indicative of competence in everyday prospective memory tasks outside the laboratory. This issue of validity will be investigated by testing the same subjects in a variety of short- and long-term prospective memory tasks, both inside and outside the laboratory, together with detailed diary studies of prospective memory failures. Maylor (1990a) observed a significant (albeit weak) relationship between performance on two nonlaboratory prospective memory tasks, but this was possibly due to the use of a common strategy. The use of cues will therefore be prohibited or controlled for in the analyses.
Finally, age comparisons across different prospective memory tasks remain to be performed. It has been suggested that age effects should be particularly evident in time-based prospective memory tasks in which environmental support is low and self-initiated activity (e.g., clock-monitoring) is high. Age deficits should be less apparent in event- and activity-based tasks (greater environmental support available and less self-initiated activity required). Also, we need to explore possible interactions between ageing and the cognitive demands of the ongoing activity that has to be interrupted to fulfil the requirements of the prospective memory task.
C. Analysis of Performance Distributions
In a recent article (3.66), Maylor & Rabbitt described the application of Brinley plots to performance distributions from two speeded tasks. Response times within each of two age groups were ranked and then plotted against each other, so that the best younger adult was paired with the best older adult, and so on. For both tasks, linear fits to the functions were almost perfect, with slopes greater than 1 and negative intercepts; these parameters were shown to be significantly different in the two tasks. First, it should be noted that although the data were cross-sectional, they were at least consistent with the possibility that slower adults are more affected by ageing than faster adults in absolute terms but not in relative terms (c.f. Maylor & Rabbitt, 1993, for a similar conclusion with regard to alcohol). Second, this novel approach demonstrates the use of within-group variability to reveal between-task differences in the effects of ageing. Thus far, the method has been applied only to speeded tasks, namely, letter coding and visual search. However, there is clearly potential to explore its applicability to non-speeded memory tasks, with the possibility of observing dissociations with age that have not necessarily been found in analyses based on group means (for example, between verbal and spatial material, and between memory for location and identity).
D. Dual Tasks
Age deficits in working memory are well-documented. Postural stability is also known to decline significantly in old age. Preliminary findings indicate that cognitive tasks which require particular components of working memory produce a more detrimental effect on standing balance in the elderly than in the young; further studies are planned to investigate these interactions and their possible consequences in more detail (see Perception and Action programme).
PROSPECTIVE MEMORY (Sellen)
A. Technology in Support of Human Memory
Over the past two years, Sellen has worked in collaboration with Rank Xerox EuroPARC, Cambridge, on the design of technological aids for supporting memory in the workplace. EuroPARC has in place a system of "Active Badges" — lightweight, wearable devices which emit an infra-red signal, picked up by sensors placed around the building. When people wear these badges, the system can automatically track and record people's activities, and send them a personal electronic "diary". The designers of this system (see Lamming & Newman, 1992) were hopeful that this would form the foundation of a memory "prosthesis", helping to support people's autobiographical memory.
In order to assess the utility of the system, Sellen collaborated on a diary/ questionnaire study of everyday memory problems in order to document the range of memory problems that occur in the workplace, and to assess their frequency and severity (3.148). One of the most striking findings from this study was the high percentage of "prospective" memory problems which occurred — problems of forgetting to carry out intentions. In response to this, Sellen designed and implemented a prototype reminding system to run on devices called PARC "Tabs". Tabs are more sophisticated versions of Badges, having touch-sensitive screens, and the capability of both transmitting and receiving information. This protoype is currently being implemented and integrated into the larger system, now called "Forget Me Not" (described in 3.57).
B. Memory for Intentions(Sellen, Wilkins)
Theoretical aspects of prospective memory were investigated in an experiment that made use of the Active Badge technology (3.100). Subjects' prospective memory performance was automatically recorded over the course of two weeks, in the context of their everyday work, by asking them to remember to press the button on their badge either every two hours (time task) or whenever they entered a particular room (place task). The results provided evidence for two different kinds of cognitive mechanism underlying prospective memory: internal, self prompting, and external, contextual cueing. In addition, the electronic diaries revealed that making transitions between different physical locations gave rise to more thoughts about the to-be-performed tasks than settling in any one location. This experiment underlines the importance of examining prospective memory in real world contexts, and shows how new technology can enable the systematic study of these kinds of memory phenomena.
FUTURE PROPOSALS
A. Technology in Support of Human Memory
The integration of the reminding prototype into the "Forget Me Not" system at EuroPARC marks the end of Sellen's involvement in this project. The details are now in the hands of the implementors. In addition, the technological infrastructure necessary to support this system is currently being installed in the Computer Science Department at the University of Toronto. Prof. Bill Buxton has suggested that this also serve as a testing site for the system, and that it be integrated and, if necessary, redesigned, as a result of its use there.
B. Memory for Intentions
Using the technology in place at EuroPARC, further experiments will investigate the cognitive processes underlying memory for intentions. Of particular interest is the frequency with which thoughts about intentions are brought to mind in time-based remembering tasks. With no specific external cue available to trigger the remembering of time-based intentions, it is unclear how remembering on the basis of time occurs. The work so far has suggested that the frequency of thoughts in time-based tasks, is, to some extent, under an individual's control, and that self-prompting is necessary precisely because externally imposed cues are not available.
To test the hypothesis that reliance on an external reminder decreases the amount of internal, self-prompting necessary, we plan an experiment similar to Sellen et. al (3.100) in which subjects at EuroPARC are again assigned the task of pressing a button on their Active Badges at pre-specified time intervals. This time, however, they will be given audio reminders either at the moment when they should perform the task, or at various time intervals before this moment, during the course of their working week. In a different week, subjects will be given no reminder at all. We will analyse the frequency of reported thoughts about the Time task, expecting to observe an increased level of self-initiated thoughts in the "no reminder" condition. The distribution of thoughts according to the variable interval for the time reminders will also be informative about the relationship between reminders and intentions. In addition, the Time task will be designed to be performed at more frequent intervals, so that we can gather more data than in the first experiment. For example, we would like to know if worrying or thinking about the task actually improves prospective memory performance.
MEMORY FOR PEOPLE (A Young)
Introduction
As A. Young joined the APU in September 1993, this report is largely based on 6 months work. A principal focus has been disorders of face memory after brain injury (Young, 1992), and particularly the extent to which prosopagnosia (neurologically-based inability to recognise familiar faces) can be considered a highly selective impairment of face memory.
A fundamental issue in visual cognition is whether the stored representations describing the appearances of known visual stimuli are separate from the representations used in perceiving those stimuli. For approaches such as Marr's (1982), recognition is a two-stage process in which a percept is created to describe a seen object's three-dimensional structure and then matched against a catalogue of the structures of known objects. This distinction between the visual percept and the stored representation of a familiar object is elided in some of the presently fashionable connectionist approaches, which emphasise the point that there are circumstances in which same network can be used in responding to novel stimuli and to those already learnt (McClelland & Rumelhart, 1988).
A. Neuropsychological Impairments(Young, de Haan, Hellawell, Humphreys, Riddoch)
Impairments of familiar face recognition provide a useful empirical means for examining these competing theoretical claims. At present, many researchers think that there is an underlying difference between cases in which prosopagnosia reflects a perceptual impairment which is sufficiently severe to prevent recognition, and cases in which there is impaired face memory, in the sense of loss of stored knowledge of the appearances of familiar people (De Renzi, Faglioni, Grossi & Nichelli, 1991; McNeil & Warrington, 1991). Others, however, have argued that this perceptual vs. mnestic distinction is actually an idealisation of a unitary problem that can vary in degree of severity, with the apparently perceptual cases simply being the more severely impaired (Levine & Calvanio, 1989).
One key to whether there are qualitatively different types of face recognition impairment may lie in face imagery. It is widely believed that imagery requires access to the same stored knowledge of appearances as is needed for recognition, from which it would follow that imagery abilities can be used to test access to this stored knowledge. A study has therefore investigated face imagery for two patients already well-documented in the research literature, HJA and PH, who experience profound difficulties in recognising familiar faces (3.93). For all face imagery tasks, PH's overall performance was severely impaired. HJA, though, showed preserved face imagery when imaging single faces and when making feature-based comparisons between imaged faces.
The interest of these findings lies in the fact that, although HJA's imagery abilities are better preserved than PH's, it is HJA who has the more widespread and severe recognition impairment. HJA has a perceptual impairment that compromises the integration of features into a coherent representation, leaving him unable to recognise many everyday objects as well as familiar faces, and he has not shown any evidence of covert recognition of faces in indirect tests (Humphreys, Troscianko, Riddoch, Boucart, Donnelly & Harding, 1992). PH, by contrast, does not experience any obvious recognition problems for most everyday objects in his daily life, and has been found to show extensive covert recognition effects with familiar faces (de Haan, Young & Newcombe, 1987; Young & de Haan, 1992).
These observations are important to the debate as to whether prosopagnosia is a unitary or a multi-stage disorder. If both patients' visual recognition difficulties are due to the same form of underlying problem (albeit differing in severity), then one must predict that, since HJA's visual recognition problems are more severe than PH's, HJA's face imagery will also be the more severely impaired. Since this is the reverse of what has been found, the results create severe problems for any account assuming a single underlying deficit. In contrast, they can be explained quite simply if it is accepted that impairments of face recognition can have different causes.
Further work in the last 6 months has involved reviews of the literature on face recognition (Young, 1994b) and the relation between recognition mechanisms and awareness (Young, 1994a; Young, 1994c). This latter line of work arises from observations of preserved priming effects from faces which are not consciously recognised in cases of prosopagnosia (de Haan, Young & Newcombe, 1991a). These findings of covert recognition in prosopagnosia show a striking parallel to phenomena of implicit memory in cases of amnesia (Schacter, 1987). The existence of such parallels between prosopagnosia and amnesia has been one of the main reasons for considering some cases of prosopagnosia as a domain-specific memory impairment.
B. Experiments with Normal Subjects (Young, Ellis, Flude, Hellawell)
A series of experiments examining the nature of semantic priming effects in the recognition of familiar people has been completed (3.92). Priming effects have provided useful ways of examining the organisation of mechanisms involved in the recognition of familiar people (Ellis, 1992), and they are also turning out to be useful in neuropsychological work (Burton, Young, Bruce, Johnston & Ellis, 1991). In particular, comparisons of repetition and semantic priming have been instructive.
Repetition priming involves the facilitation of recognition by a previously recognised stimulus with the same identity. For example, recognition of the comedian Mel Smith's face is faster if his face has appeared previously in the experiment than if it has not come up before (Bruce & Valentine, 1985). Repetition priming effects are long-lasting (being found across intervals of several minutes in existing published studies, and as long as three months in as yet unpublished work of our own) and domain-specific (except at very short time intervals, recognition of Mel Smith's face is not facilitated by a recent encounter with the name 'Mel Smith').
Semantic priming involves the facilitation of recognition by a previously recognised stimulus with a different, but semantically related, identity. For example, recognition of Mel Smith's face may be faster if the face of Griff Rhys-Jones (who often performs with him) has appeared previously in the experiment (Bruce & Valentine, 1986). In contrast to repetition priming effects, the facilitation produced by semantic priming is very short-lived (generally dissipating within seconds), yet can cross stimulus domains (for example, from recognition of Mel Smith's face to recognition of Griff Rhys-Jones's name).
These differences between repetition priming and semantic priming effects are widely taken to indicate that the sources of facilitation arise at different loci in the recognition system, but more precise specification of the underlying reasons for the differences has proved difficult. However, a considerable advance has been made with the development of an interactive activation implementation of a variant of the Bruce and Young (1986) model of face recognition by Burton, Bruce and Johnston (1991). This interactive activation model is able to simulate the effects of repetition and semantic priming, and the differences between them.
Experiments have shown that repetition and semantic priming have different loci in the recognition system, and that there is no facilitation of decisions for which the identity of the stimulus does not need to be accessed (3.92). We also demonstrated that an important contribution to semantic priming effects is made by associative relatedness (recognition of Eric Morecambe's face is facilitated by having just seen Ernie Wise), rather than category membership per se (recognition of Eric Morecambe's face is not facilitated by having just seen any comedian other than Ernie Wise), and extended this observation by comparing associative priming and priming from the 'same person' in a cross-domain (face prime, name target) paradigm. These results are consistent with the interactive activation simulation developed by Burton et al., (1991), and set constraints which will have to be met by any other plausible account of semantic priming.
FUTURE PROPOSALS
Introduction
Future work aims to explore retrograde and anterograde components of face memory, and their relation to semantic memory for people. As in past work, a combination of investigations of normal people and clinical cases will be employed. Work with neuropsychological cases must always be somewhat opportunistic, since it depends on the availability and willingness of suitable volunteers; this has not been a major problem in the past, despite the rarity of some of the conditions investigated.
Investigations of face processing present an important opportunity to study some of our most highly developed visual abilities with stimuli that are rich in social meaning. The apparent ease with which we recognise faces can, however, be deceptive. In fact, there are many occasions when misidentifications and other forms of error can happen (Young, 1993). The starting point for planned future work is thus the observation that failures of recognition can be quite orderly, which applies in everyday life (Young, Hay & Ellis, 1985) and for recognition failures in the laboratory (Hay, Young & Ellis, 1991). More recently, converging evidence from neuropsychological case studies (de Haan, Young & Newcombe, 1991b; Flude, Ellis & Kay, 1989; Young, 1992) indicates that the functional organisation of the face recognition system involves sequential access to different types of information. Recent theoretical work has therefore concentrated on providing a simulation of how this is might be achieved (Burton & Bruce, 1993).
Although all of this represents useful progress, there are some noticeable rough edges requiring more detailed attention. For example, it has often been noted that much of the functional modelling enterprise has ignored issues of learning. Recent simulation work has sought to correct this oversight (Burton, in press), but learning remains under-investigated in neuropsychology. A detailed examination of retrograde and anterograde components of face memory after brain injury will therefore be useful.
A. Neuropsychological Impairments (Young)
Particularly relevant neuropsychological conditions include prosopagnosia, and what Ross (1980) calls visual recent memory loss.
A1. Anterograde Impairments in Prosopagnosia, and Visual Recent Memory Loss: It is widely held that some cases of prosopagnosia reflect impaired face memory, in the sense of damage to pre-morbidly acquired representations of the appearances of familiar people. If so, they usually involve a combination of dense retrograde and anterograde impairments. In many published reports, the retrograde loss is so severe that hardly any premorbidly familiar faces can be recognised overtly, and performance on anterograde face memory tests, such as the Faces part of the Warrington Recognition Memory Test (Warrington, 1984), is at chance.
In some types of simulation (Burton, in press), this combination of retrograde and anterograde impairments might well be considered unsurprising, because both defects would arise from a common locus of damage. However, it is also possible to find cases of impaired anterograde memory for faces in which there is no retrograde loss, and premorbidly familiar faces are recognised without difficulty. Damasio and his colleagues refer to these as 'anterograde prosopagnosias' (Tranel & Damasio, 1985), whereas others prefer the more clumsy but neutral term 'visual recent memory loss' (Ross, 1980).
These cases of impaired anterograde memory for faces are important theoretically, because they imply that full prosopagnosia might reflect a combination of separable anterograde and retrograde impairments. This would constrain theories of the type of learning mechanism involved, but it would depend on a convincing demonstration that the anterograde impairments found in prosopagnosia and visual recent memory loss arise for the same reasons. Thus both conditions require more detailed investigation.
An additional reason for interest in anterograde memory for faces is that, for a thoroughly studied case of loss of visual recent memory (Hanley, Pearson & Young, 1990; Hanley, Young & Pearson, 1991), the defect was found to coincide with impairment of the visuo-spatial sketchpad. This would be predicted by Baddeley's (1986) working memory model, in which the visuo-spatial sketchpad plays a role in new visual learning.
Further single-case studies of visual recent memory loss are therefore planned, to determine whether this link to impairment of the visuo-spatial sketch pad forms a common feature, to explore in more detail the relation of the problem to the anterograde memory deficit found in cases of prosopagnosia, and to determine whether any involve face-specific memory deficits. For the latter purpose, parallel versions of Warrington's Recognition Memory Test (Warrington, 1984) are being developed, involving a wider range of stimulus materials. Of course, one cannot ever prove that an anterograde impairment is confined exclusively to stimulus class x, because it is always possible to think of some stimulus class y which happened not to be tested. But the tactic of testing a range of materials should be useful. At present, it is known that patient ELD (Hanley et al., 1990; Hanley et al., 1991), was poor at learning other new visual materials as well as faces, and this is consistent with an account in terms of impairment of the visuo-spatial sketchpad. It will be important to discover whether more selective cases exist.
A2. Retrograde Impairments in Prosopagnosia and Semantic Memory Loss: An obvious implication of the hypothesis that prosopagnosia is a combination of separable anterograde and retrograde face memory impairments is that, as well as Damasio's anterograde prosopagnosias, a purely retrograde form might exist. Although this has never been reported, cases of gradual recovery from prosopagnosia have occasionally been noted. This gradual recovery would be consistent with an impairment affecting only retrograde face memory, though it can also be explained in other ways. The crucial test will be to determine whether recovery extends to all premorbidly familiar faces, or only to those faces encountered postmorbidly.
A second general area concerns the relation between memory representations involving facial appearance and personal semantic characteristics. In neuropsychological studies, there is a clear contrast between prosopagnosic cases, in which recognition from the face is impaired but recognition from other cues (such as the person's name) remains relatively unaffected, and cases which reflect a more central loss of semantic information concerning familiar people, who show inability to recognise people from face, name or voice (Ellis, Young & Critchley, 1989; Hanley, Young & Pearson, 1989). This latter type of problem has not yet attracted a lot of attention, despite the increase in interest in other aspects of semantic memory. Effects of quite basic properties, such as frequency of encounter and age of initial learning, need to be explored. Development of appropriate techniques could form part of a more general investigation of recent claims that age of acquisition affects the naming of seen objects and words but not their recognition (Morrison, Ellis & Quinlan, 1992). Faces and people's names form useful stimuli for this purpose, since new examples are being learnt throughout life.
B. Experiments with Normal Subjects (Young, Calder)
B1. Priming Effects: Part of the reason for the relatively scant investigation of neuropsychological cases involving loss of semantic information concerning familiar people lies in the lack of suitable methods which can be imported from experimental work. Effort will therefore be devoted to research on normal people which could provide the basis for such techniques. This would include further investigation of priming effects using predictions derived from the Burton et al. model (Burton, Bruce & Johnston, 1990) and possible alternatives (Brédart, 1993; Valentine, Moore, Flude, Young & Ellis, 1993). Particular attention will be given to the phenomenon of self priming (3.92) which involves cross-domain facilitation of recognition of the same person (for example, seeing John Major's face primes recognition of the target name 'John Major'). This effect is particularly useful in neuropsychological work, and it provides an interesting test of the Burton et al. model (Burton et al., 1990), which predicts its time course and other properties.
B2. Mapping of Visual and Semantic Descriptions of Familiar People: An important question in face recognition is whether the visual representations used to recognise familiar faces are abstract descriptions of the essential properties of a particular face's structure, or are better considered as accretions of individually encountered instances in which the face was seen (Young & Bruce, 1991). An equivalent issue arises concerning the nature of semantic representations, which are treated in a highly abstract way in some current models of person recognition (Burton et al., 1990). For this reason, the extent to which semantic representations of familiar people remain linked to, or become independent of, the circumstances in which they were learnt will be investigated. For example, people's appearances change markedly throughout their lives, yet it is not known whether access to a fact about a person learnt when s/he was aged 20 (for example, that Cliff Richard sang 'Living Doll') will be more readily cued by a contemporary picture (Cliff with his quiff) than a picture of the same face from a different era (Cliff Richard now). Abstractive models predict no difference in matching Cliff Richard's face at age 20 and his face at age 50 to the facts that he sang Living Doll and now wants to be Heathcliff. In contrast, instance-based models predict a crossover interaction: for Cliff Richard's face at age 20, verifying that he sang Living Doll should be faster than verifying that he wants to be in Wuthering Heights; for Cliff Richard's face at age 50, the reverse would apply.
DISORDERS OF SEMANTIC MEMORY (K Patterson)
Introduction
Semantic memory is psychology's term for the component of long-term memory that represents knowledge rather than memories of specific events or episodes. The goal of this research is to advance our understanding of semantic memory by studying its deterioration in two types of neurodegenerative disease: semantic dementia, which appears to be a distinct syndrome selectively affecting semantic memory, and dementia of the Alzheimer's type (DAT). Although my principal interest in semantic memory disorders relates to their impact on language processes (see K Patterson's section in the Language and Communication Programme), certain issues regarding semantic memory per se are (as a result of a collaboration with Dr J R Hodges in the Neurology Department) becoming part of my research programme. We are using evidence from the breakdown of semantic memory to ask such questions as (i) how semantic memory, as a whole 'module' or as sub-systems defined by different types of knowledge, is neuroanatomically localised in the brain; (ii) whether semantic and episodic memory are truly separable memory systems; and (iii) how semantic information is organised with regard to different types of knowledge, e.g. for objects as compared to words, for information relating to different sensory systems, etc.
A. Neuroanatomy of Semantic Memory
A1. Semantic Dementia: To date, few identified cases of semantic dementia have come to post mortem; but in all of these, the affected areas have been predominantly in the temporal lobes (Hodges, 1993; Snowden et al, 1992). Structural (MRI) and functional (PET or SPET) in vivo imaging of patients with semantic dementia have also shown abnormalities almost exclusively in temporal neocortex, either predominantly in the left hemisphere or sometimes bilaterally (3.55). These findings, together with a review of other literature on disorders of semantic memory, have led us to conclude that lateral areas of temporal cortex seem to be crucial to the operation of semantic memory (3.140).
A2. DAT: In DAT, the pathological changes characteristic of the disease begin typically around the medial temporal structures critical for episodic rather than semantic memory (Braak & Braak, 1991); thus the earliest and most profound deficit in DAT is one of episodic memory. As the disease progresses, however, pathology spreads to bilateral posterior association cortex, including the lateral temporal structures that are more selectively affected in semantic dementia. Disrupted semantic memory -- which may be relatively mild, though detectable, in early stages of DAT (3.97) -- then becomes a major feature in later stages (Hodges, Salmon & Butters, 1992).
B. The Episodic/Semantic Distinction
B1. Conceptual Issues: The separability of episodic and semantic memory 'systems' is by no means established. The fact that you know that Paris is the capital of France without being able to retrieve any specific memories of when or where you acquired this knowledge may seem to dissociate semantic from episodic memory; but it might instead only reflect the accumulation of many learning episodes in many different contexts over many years. On the basis of the neuropsychological double dissociation between patients with the amnesic syndrome (who have profoundly disrupted episodic memory with sometimes well preserved semantic memory) and patients with semantic dementia (who have profoundly disrupted semantic memory with apparently well preserved episodic memory), our current working hypothesis is that semantic and episodic memory are in fact separable systems with neuroanatomically and psychologically distinct bases (3.140). The operation of these two components of long-term memory, however, is probably interdependent, in two senses. First, since learning new information necessitates some memory of the event conveying that information, no new semantic memories will be acquired if the brain mechanisms for maintaining episodes are severely compromised. Secondly, since information that is poorly understood is difficult to encode in a durable fashion, no new episodic memories will be acquired if the brain mechanisms for comprehending information are severely compromised.
B2. Empirical Evidence: We have only begun to tackle the problem of assessing the status of episodic memory in patients with semantic dementia. On the Autobiographical Memory Interview (Kopelman et al, 1989), several patients achieved performance within normal limits when relating autobiographical episodes, but were impaired when asked to give personal semantic information (3.55). Kim Graham's thesis research has established that a patient with semantic dementia can perform normally on a recognition memory test (episodic memory) for items from a previously administered semantic memory assessment on which his performance was markedly impaired.
C. Organisation of Semantic Memory
C1. Is Knowledge Organised Hierarchically? In our initial investigations of the organisation of semantic memory, as revealed by the profile of its deterioration in semantic dementia, we have completed two intensive longitudinal single-case studies, each of about 3 years' duration, ceasing at the point when each patient was too impaired to be testable (3.54, 3.96). One major issue is the hypothesis (Warrington, 1975) that concept knowledge is organised hierarchically, with general superordinate knowledge (e.g. that an object is a living creature) at the top of the hierarchy, exemplar knowledge (e.g., of dogs and horses as members of the class of land mammals) in the middle, and very specific attribute knowledge (e.g., dogs eat meat and bark, horses are herbivorous and neigh) at the bottom. Like Warrington's (1975) cases, our two progressive patients have shown a dramatic contrast between relatively preserved performance on tests tapping the more general superordinate aspects of knowledge and marked impairment on the more specific levels, both across tests at a given point in time and in their longitudinal pattern of decline. The interpretation of this contrast is, however, debatable. Our current hypothesis is that it does not imply anything hierarchical about the organisation of semantic knowledge, but rather reflects the natural consequences of loss of specific features from a distributed network of featural knowledge about objects (3.96, 3.140; see also Rapp & Caramazza, 1993). A reduced set of features will support 'higher-level' distinctions, such as that between living and manmade objects, but not more specific ones concerning attributes such as barking/neighing.
C2. Is Knowledge Organised by Category and/or Modality? The literature of the last decade abounds with tantalising descriptions of highly selective disorders of semantic memory (mainly in cases of herpes simplex encephalitis, though also in a few cases of either cerebrovascular accident or head injury). Some disorders are apparently category specific, with a marked contrast between the patient's performance on living things and on man-made objects; typically the superior performance is on objects, but there are just enough reports of an advantage with living things for theorists to propose a double dissociation. Some disorders are apparently modality specific, with a marked contrast between comprehension of objects and of words. Indeed, a few cases showing both effects have led McCarthy (1994) to argue that semantic memory may have four basic quadrants defined by these two primary distinctions between category (natural kind/artefact) and modality (visual/verbal). Interestingly, we have observed only hints of these two distinctions in our cases of either semantic dementia or DAT; and those hints have been in directions explicable on other grounds. For example, better discrimination between living and manmade objects from pictures than from words is understandable because pictures offer structure/function correlations not obvious from word forms.
FUTURE PROPOSALS
A. Neuroanatomy of Semantic Memory
A1. Semantic Dementia: In terms of leading research groups working on this disorder, we are probably one of two in the UK (the other being Drs D Neary and J Snowden at the Manchester Royal Infirmary), and one of four or five in the world. Although semantic dementia is a relatively rare condition, John Hodges is likely to pick up any putative case in the East Anglian Region; and, as our work on this topic is becoming known, relevant patients from other regions are being referred to us. We plan serial in vivo imaging (MRI and SPET) in our longitudinal case studies; with the advent of PET (and possibly functional MRI) in Cambridge, the quality of functional brain imaging will soon be substantially improved. Most patients and their families also agree to donate brain tissue at death and, owing to cooperation with the Cambridge Brain Bank (Dr John Xuereb), we can obtain post-mortem pathological analysis. We therefore have the opportunity, over the next 5 years or so, to make major advances concerning brain-behaviour relationships for semantic memory. Although the imaging and pathology clearly rely on the Addenbrooke's side of the collaboration, the cognitive neuropsychological expertise offered by the APU is an equally vital component of the brain-behaviour equation. For example, in order to determine whether visual and verbal semantic knowledge are neuroanatomically separable, it is essential to have tests (e.g., 3.156) where performance measuring access to knowledge from either pictures or words can be correlated with neuroanatomical data.
A2. DAT: In our current 3-year longitudinal study, for a subset of our cohort of about 50 DAT patients, we have both imaging data and declarations of intent concerning post-mortem brain tissue; and in the planned sequel to this study, imaging will play a more prominent role. One of the central questions to be addressed is the correlation between degree of semantic memory impairment (which is highly variable at mild-to-moderate stages of DAT) and structural/ functional abnormalities of the temporal-lobe structures that, by our hypothesis, are critical for semantic memory.
B. The Episodic/Semantic Distinction
B1. Semantic Dementia: Patients with this syndrome seem ideal for addressing the controversial question of whether these two components of long-term memory really differ in kind or only in degree. If the apparent distinction only reflects the fact that knowledge is the accumulation of many episodes, then semantic memory should be more resistant to brain damage, and the advantage for semantic > episodic memory in patients with the amnesic syndrome would not constitute dramatic evidence for a real dissociation. If, on the other hand, we can demonstrate a significant advantage for episodic > semantic memory in patients with semantic dementia, the case for separability will be stronger. In addition to the neuroanatomical approach to this issue described above, at a purely behavioural level, we plan to expand the two lines of investigation already started (the Autobiographical Memory Inventory (Kopelman et al, 1989), and assessment of patients' episodic memory for recently administered tests of semantic memory; see B2 above), and also to develop some new lines such as Mandler's (1990) technique for assessing episodic memory in preverbal normal infants: the infant observes the experimenter performing a distinctive sequence of actions with a small set of objects and is later given the opportunity to re-enact this sequence and thereby display memory for the event. As this test depends neither upon knowledge of the identity of the objects nor upon comprehension of verbal instructions, we hope that these procedures can be adapted for use with patients at advanced stages of semantic dementia.
C. Organisation of Semantic Memory
C1. Hierarchical and Categorical Organisation: We aim to establish that patients with disrupted semantic memory do consistently achieve greater success on tests or items that tap higher-level, more general knowledge, using both a variety of stimulus materials (words, pictures, objects) and a variety of test paradigms (answering questions, sorting pictures or words, priming). It will be essential to demonstrate that this predicted pattern is not just attributable to differential comprehension of the terms applied to the different levels (Funnell, 1993). For example, we have very recently begun to use sorting tests where, at levels below the discrimination between "living" vs. "manmade" things, we use high-frequency category labels rather than such potentially uncommon terms as "native" vs. "foreign" animal. Preliminary results are encouraging, in the sense that two new semantic dementia patients have shown poor performance in sorting pictures of various instances and breeds into groups labelled "dog" and "cat" (extremely common terms) at the same time as performing well on the higher-level discrimination between living and manmade objects.
We are also, once again, taking inspiration from Jean Mandler's work on conceptual development in normal babies, which suggests that the distinction between living and non-living things is learned very early in development (e.g., Mandler, 1991). This is important both theoretically, for evaluating the possibility that early-acquired knowledge is represented in the brain in a way that makes it more resistant to brain-injury, and also practically, for the application of techniques designed for pre-verbal infants to the assessment of severely impaired patients. One approach involves a habituation technique, in which the subject is handed a series of five toy objects to examine: the first four are always from the same category; on some trials, the fifth in the series is also a member of that category; on other trials it is an instance of a different category. Very young infants look significantly longer at items in position 5 that cross the animal/ object boundary but show little or no discrimination between different types of animals. We are currently testing a pilot version of this procedure with patients who are too impaired to comply with the requirements of explicit sorting or matching tests.
C2. Is Knowledge Organised by Category and/or Modality? Our hypothesis is that these distinctions will eventually be explicable in terms of (a) differences in the manner in which knowledge is originally acquired and (b) differences in the procedures by which information from objects and words is retrieved or computed. We have some tentative plans about how to explore these controversial issues, but cannot guarantee to accomplish them in the near future since none of our current cohort of patients shows either of these distinctions in a prominent form.
C3. Item-Specific Consistency: Consistency can be assessed over time on the same test, where extreme consistency would correspond to the following pattern: there would be a point in the progression of a deficit, before which the patient consistently succeeds (for example) in naming a picture of a rhinoceros, and then, having once failed to name a rhinoceros, never succeeds again. It can also be measured across different tests within one time band, where extreme consistency would mean, for example, that failure to "point to the rhinoceros" in a set of animal pictures should predict that the patient will also fail when asked to name the picture, to answer questions such as "Does a rhinoceros have a horn?", etc. There are substantial statistical problems in measuring both of these forms of consistency, owing to declining baselines over time in the first case and differing levels of difficulty (and of chance performance) for the various tests in the second case. There are also substantial problems in interpreting statistically reliable degrees of consistency, since -- when considering item-specific consistency between two tests, for example -- performance on both may be strongly modulated by some other factor such as word frequency. We will require considerable statistical assistance from Dr Ian Nimmo-Smith in working out solutions to these measurement/interpretation problems. The goal is a worthy one, however, since the issue of consistency plays a major role in theoretical discussions of semantic disorders, particularly in attempts to identify specific disorders as deficits of either "storage" (i.e., genuine loss of information from the brain system representing that knowledge) or "access" (problems in activating or retrieving knowledge that is still somehow coded in the system) (see Rapp & Caramazza, 1993 and Shallice, 1988, for discussion).
PUBLICATIONS (Excluding work done prior to arrival at APU)
Authored Books
3.1. BADDELEY, A.D. (1990). Human Memory: Theory and Practice. London: Lawrence Erlbaum Associates.
3.2. BADDELEY, A.D. (1993). Your Memory: A User's Guide (2nd edition). London: Lifecycle Publications.
3.3. Gathercole, S.E. & BADDELEY, A.D. (1993). Working Memory and Language. Hove, Sussex: Lawrence Erlbaum Associates.
Edited Books
3.4. Broadbent, D.E., BADDELEY, A.D. & Reason, J.T. (Eds.) (1990). Human Factors in Hazardous Situations. Oxford: Clarendon Press.
3.5. BADDELEY, A. & Weiskrantz, L. (Eds.) (1993). Attention: Selection, Awareness and Control. A Tribute to Donald Broadbent. Oxford: Clarendon Press.
3.6.* BADDELEY, A.D., WILSON, B.A., & WATTS, F. (Eds.) (in press). Handbook of Memory Disorders. Chichester: John Wiley & Sons.
Refereed Articles
3.7. Alberoni, M., BADDELEY, A.D., Della Sala, S., Logie, R. & Spinnler, H. (1992). Keeping track of a conversation: Impairments in Alzheimer's disease. International Journal of Geriatric Psychiatry, 7, 639-646.
3.8. ANDRADE, J. (in press). Learning during anaesthesia: A review. British Journal of Psychology.
3.9. ANDRADE, J. & Meudell, P.R. (1993). Is spatial information encoded automatically in memory? Quarterly Journal of Experimental Psychology, 46A, 365-375.
3.10. ANDRADE, J. & Munglani, R. (in press). Are therapeutic suggestions really therapeutic? (letter to the editor), British Journal of Anaesthesia.
3.11. ANDRADE, J., Munglani, R., Jones, J.G. & BADDELEY, A.D. (1994). Cognitive performance during anaesthesia. Consciousness and Cognition, 3, 148-165.
3.12. BADDELEY, A.D. (1992). Is working memory working? The Fifteenth Bartlett Lecture. Quarterly Journal of Experimental Psychology, 44A, 1-31.
3.13. BADDELEY, A.D. (1992). Working memory. Science, 255, 556-559
3.14. BADDELEY, A.D. (1992). Working memory: The interface between memory and cognition. Journal of Cognitive Neuroscience, 4, 281-288.
3.15. BADDELEY, A.D. (1993). A theory of rehabilitation without a model of learning is a vehicle without an engine: A comment on Caramazza and Hillis. Neuropsychological Rehabilitation, 3, 235-244.
3.16. BADDELEY, A.D. (1993). Short-term phonological memory and long-term learning: A single case study. European Journal of Cognitive Psychology, 5, 129-148.
3.17. BADDELEY, A.D. (1994). The magical number seven: Still magic after all these years? Psychological Review, 101, 353-356.
3.18. BADDELEY, A.D. & Hitch, G.J. (in press). Developments in the concept of working memory. Neuropsychology.
3.19. BADDELEY, A.D. & ANDRADE, J. (in press). Reversing the word length effect: A comment on Caplan, Rochon and Walters. Quarterly Journal of Experimental Psychology.
3.20. BADDELEY, A.D., Bressi, S., Della Sala, S., Logie, R. & Spinnler, H. (1991). The decline of working memory in Alzheimer's Disease: A longitudinal study. Brain, 114, 2521-2542.
3.21. BADDELEY, A.D., Della Sala, S. & Spinnler, H. (1991). The two-component of memory deficit in Alzheimer's Disease. Journal of Clinical and Experimental Neuropsychology, 13, 373-380.
3.22. BADDELEY, A.D., EMSLIE, H. & NIMMO-SMITH, I. (1993). The Spot-the-Word test: A robust estimate of verbal intelligence based on lexical decision. British Journal of Clinical Psychology, 32, 55-65.
3.23. BADDELEY, A.D. & Hitch, G.J. (1993). The recency effect: Implicit learning with explicit retrieval? Memory and Cognition, 21, 146-155.
3.24. BADDELEY, A.D. & Hitch, G.J. (in press). Developments in the concept of working memory. Neuropsychology.
3.25. BADDELEY, A.D., Meeks Gardner, J. & Grantham-McGregor, S. (in press). Cross-cultural cognition: Developing tests for developing countries. Applied Cognitive Psychology.
3.26. BADDELEY, A.D., PAPAGNO, C. & ANDRADE, J. (1993). The sandwich effect: The role of attentional factors in serial recall. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 862-870.
3.27. BADDELEY, A.D. & WILSON, B.A. (1993). A developmental deficit in short-term phonological memory: Implications for language and reading. Memory, 1, 65-78.
3.28. BADDELEY, A.D. & WILSON, B.A. (1994). A case of word deafness with preserved span: Implications for the structure and function of short-term memory. Cortex, 29, 741-748.
3.29. BADDELEY, A.D. & WILSON, B.A. (1994). When implicit learning fails: Amnesia and the problem of error elimination. Neuropsychologia, 32, 53-68.
3.30. BEKERIAN, D.A. (1993). In search of the typical eyewitness. American Psychologist, 48, 574-576.
3.31. BEKERIAN, D.A. (in press). The problems with childhood memories. Consciousness and Cognition: Special Edition on Recovered/False Memories.
3.32. BEKERIAN, D.A. & DENNETT, J.L. (1990). Spoken and written recall of visual narratives. Applied Cognitive Psychology, 4, 175-187.
3.33. BEKERIAN, D.A. & DENNETT, J.L. (1993). The cognitive interview technique: Reviving the issues. Applied Cognitive Psychology, 7, 275-297.
3.34. Bradley, B.P. & BADDELEY, A.D. (1990). Emotional factors in forgetting. Psychological Medicine, 20, 351-355.
3.35. CLARE, L., McKenna, P.J., Mortimer, A.M. & BADDELEY, A.D. (1993). Memory in schizophrenia: What is impaired and what is preserved? Neuropsychologia, 31, 1225-1241.
3.36. Cockburn, J., WILSON, B., BADDELEY, A. & Hiorns, R. (1990). Assessing everyday memory in patients with dysphasia. British Journal of Clinical Psychology, 29, 353-360.
3.37. Cockburn, J., WILSON, B.A., BADDELEY, A.D. & Hiorns, R. (1990). Assessing everyday memory in patients with perceptual deficits. Clinical Rehabilitation, 4, 129-135.
3.38. Cohen, G., Conway, M.A. & MAYLOR, E.A. (in press). Flashbulb memories in older adults. Psychology and Aging.
3.39. da Costa Pinto, A. & BADDELEY, A.D. (1991). Where did you park your car? Analysis of a naturalistic long-term recency effect. European Journal of Cognitive Psychology, 3, 297-313.
3.40. DE WALL, C., WILSON, B.A. & BADDELEY, A.D. (1994). The Extended Rivermead Behavioural Memory Test: A measure of everyday memory performance in normal adults. Memory, 2, 149-166.
3.41. DRITSCHEL, B.H., WILLIAMS, J.M.G., BADDELEY, A.D. & NIMMO-SMITH, I. (1992). Autobiographical fluency: A method for the study of personal memory. Memory and Cognition, 20, 133-140.
3.42. Eldridge, M., BARNARD, P. & BEKERIAN, D. (1994). Autobiographical memory and daily schemas at work. Memory, 2, 51-74.
3.43. Gathercole, S.E. & BADDELEY, A. (1990). Phonological memory deficits in language disordered children: Is there a causal connection? Journal of Memory and Language, 29, 336-360.
3.44. Gathercole, S.E. & BADDELEY, A.D. (1990). The role of phonological memory in vocabulary acquisition: A study of young children learning new names. British Journal of Psychology, 81, 439-454.
3.45. Gathercole, S.E. & BADDELEY, A.D. (1993). Phonological working memory: A critical building block for reading development and vocabulary acquisition. European Journal of Psychology of Education, 8, 259-272.
3.46. Gathercole, S.E., WILLIS, C.S. & BADDELEY, A.D. (1991). Differentiating phonological memory and awareness of rhyme: Reading and vocabulary development in children. British Journal of Psychology, 82, 387-406.
3.47. Gathercole, S.E., WILLIS, C.S. & BADDELEY, A.D. (1991). Nonword repetition, phonological memory, and vocabulary: A reply to Snowling, Chiat, and Hulme. Applied Psycholinguistics, 12, 375-379.
3.48. Gathercole, S.E., WILLIS, C.S., BADDELEY, A.D. & EMSLIE, H. (1994). The children's test of nonword repetition: A test of phonological working memory. Memory, 2, 103-127.
3.49. Gathercole, S.E., WILLIS, C.S., EMSLIE, H. & BADDELEY, A.D. (1991). The influences of number of syllables and wordlikeness on children's repetition of nonwords. Applied Psycholinguistics, 12, 349-367.
3.50. Gathercole, S.E., WILLIS, C.S., EMSLIE, H. & BADDELEY, A.D. (1992). Phonological memory and vocabulary development during the early school years: A longitudinal study. Developmental Psychology, 28, 887-898.
3.51. GREEN, R.E.A. & SHANKS, D.R. (1993). On the existence of independent explicit and implicit learning systems: An examination of some evidence. Memory and Cognition, 21, 304-317.
3.52. Happel, B.L.M., & MURRE, J.M.J. (in press). The design and evolution of modular neural network architectures. Neural Networks.
3.53. Heemskerk, J.N.H., Hoekstra, J., MURRE, J.M.J., Kemna, L.H.J.G. & Hudson, P.T.W. (1994). The BSP400: A modular neurocomputer. Microprocessors and Microsystems, 18, 67-78.
3.54. Hodges, J., PATTERSON, K. & Tyler, L. (in press). Loss of semantic memory: Implications for the modularity of mind. Cognitive Neuropsychology.
3.55. Hodges, J.R., PATTERSON, K., Oxbury, S. & Funnell, E. (1992). Semantic dementia: Progressive fluent aphasia with temporal lobe atrophy. Brain, 115, 1783-1806.
3.56. KOLODNY, J.A. (1994). Memory processes in classification learning: An investigation of amnesic performance in categorization of dot patterns and artistic styles. Psychological Science, 5, 164-169.
3.57. Lamming, M., Brown, P., Carter, K., Eldridge, M., Flynn, M., Louie, G., Robinson, P., & SELLEN, A. J. (1994). The design of a human memory prosthesis. Computer Journal , Vol. 37, No. 3.
3.58. LEVEY, A.B. & Martin, I. (in press). Human classical conditioning: The status of the CS. Integrative Physiological and Behavioural Sciences.
3.59. Logie, R.H., Zucco, G.M., & BADDELEY, A.D. (1990). Interference with visual short-term memory. Acta Psychologica, 75, 55-74.
3.60. MACLEOD, A.K., WILLIAMS, J.M.G. & BEKERIAN, D.A. (1991). Worry is reasonable: The role of explanations in pessimism about future personal events. Journal of Abnormal Psychology, 100, 478-486.
3.61. Martin, I. & LEVEY, A.B. (1991). Blocking observed in human eyelid conditioning. Quarterly Journal of Experimental Psychology, 43B, 233-256.
3.62. MAYLOR, E.A. (1993). Aging and forgetting in prospective and retrospective memory tasks. Psychology and Aging, 8, 420-428.
3.63. MAYLOR, E.A. (1994). Ageing and the retrieval of specialised and general knowledge: Performance of Masterminds. British Journal of Psychology, 85, 105-114.
3.64. MAYLOR, E.A. (in press b). Effects of aging on the retrieval of common and proper names. Facts and Research in Gerontology.
3.65. MAYLOR, E.A. (in press c). Remembering versus knowing television theme tunes in middle-aged and elderly adults. British Journal of Psychology.
3.66. MAYLOR, E.A. & Rabbitt, P.M.A. (1994). Applying Brinley plots to individuals: Effects of aging on performance distributions in two speeded tasks. Psychology and Aging, 9, 224-230.
3.67. McKenna, P.J., Tamlyn, D., Lund, C.E., Mortimer, A.M., Hammond, S. & BADDELEY, A.D. (1990). Amnesic syndrome in schizophrenia. Psychological Medicine, 20, 967-972.
3.68. Munglani, R., ANDRADE, J., Sapsford, D.J., BADDELEY, A.D. & Jones, J.G. (1993). A measure of consciousness and memory during isoflurane administration: The coherent frequency. British Journal of Anaesthesia, 71, 633-641.
3.69. MURRE, J.M.J. (1992). From plans to mediated actions. Commentary on Bridgeman on consciousness. Psycoloquy, 3 (25), consciousness.10.
3.70. PAPAGNO, C., Valentine, T. & BADDELEY, A.D. (1991). Phonological short-term memory and foreign-language vocabulary learning. Journal of Memory and Language, 30, 331-347.
3.71. Rabbitt, P., MAYLOR, E., Stollery, B., McInnes, L., Bent, N. & Moore, B. (in press). What goods can self-assessment questionnaires deliver for cognitive gerontology? Applied Cognitive Psychology.
3.72. Salamé, P. & BADDELEY, A. (1990). The effects of irrelevant speech on immediate free recall. Bulletin of the Psychonomic Society, 28, 540-542.
3.73. Schlottmann, A. & SHANKS, D.R. (1992). Evidence for a distinction between judged and perceived causality. Quarterly Journal of Experimental Psychology, 44A, 321-342.
3.74. SHANKS, D. (1990). Connectionism and human learning: Critique of Gluck and Bower (1988). Journal of Experimental Psychology: General, 119, 101-104.
3.75. SHANKS, D. (1990). Connectionism and the learning of probabilistic concepts. The Quarterly Journal of Experimental Psychology, 42A, 209-237.
3.76. SHANKS, D.R. (1990). On the cognitive theory of conditioning. Biological Psychology, 30, 171-179.
3.77. SHANKS, D.R. (1991). Categorization by a connectionist network. Journal of Experimental Psychology: Learning, Memory, and Cognition, 17, 433-443.
3.78. SHANKS, D.R. (1991). On similarities between causal judgments in experienced and described situations. Psychological Science, 2, 341-350.
3.79. SHANKS, D.R. (1991). A connectionist account of base-rate biases in categorization. Connection Science, 3, 143-162.
3.80. SHANKS, D.R. (1992). Connectionist accounts of the inverse base-rate effect in categorization. Connection Science, 4, 3-18.
3.81. SHANKS, D.R. (1993). Human instrumental learning: A critical review of data and theory. British Journal of Psychology, 84, 319-354.
3.82. SHANKS, D.R. (1993). Associative versus contingency accounts of category learning: Reply to Melz, Cheng, Holyoak and Waldmann (1993). Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 1411-1423.
3.83. SHANKS, D. & Dickinson, A. (1990). Contingency awareness in evaluative conditioning: A comment on Baeyens, Eelen and van den Bergh. Cognition and Emotion, 4, 19-30.
3.84. SHANKS, D.R. & Dickinson, A. (1991). Instrumental judgment and performance under variations in action-outcome contingency and contiguity. Memory and Cognition, 19, 353-360.
3.85. SHANKS, D.R., GREEN, R.E.A. & KOLODNY, J. (1994). A critical examination of the evidence for unconscious (implicit) learning. In C. Umilta & M. Moscovitch (Eds.), Attention and Performance XV: Conscious and Nonconscious Information Processing. Cambridge, MA: MIT Press.
3.86. Shaw, G.A. & BEKERIAN, D.A. (1991). Hypermnesia for high-imagery words: The effects of interpolated tasks. Memory and Cognition, 19, 87-94.
3.87. Tamlyn, D., McKenna, P.J., Mortimer, A.M., Lund, C.E., Hammond, S. & BADDELEY, A.D. (1992). Memory impairment in schizophrenia: Its extent, affiliations and neuropsychological character. Psychological Medicine, 22, 101-115.
3.88. TEASDALE, J.D., Proctor, L., LLOYD, C.A. & BADDELEY, A.D. (1993). Working memory and stimulus-independent thought: Effects of memory load and presentation rate. European Journal of Cognitive Psychology, 5, 417-433.
3.89. Vallar, G., PAPAGNO, C. & BADDELEY, A.D. (1991). Long-term recency effects and phonological short-term memory: A neuropsychological case study. Cortex, 27, 323-326.
3.90. WILSON, B.A. & BADDELEY, A.D. (1993). Spontaneous recovery of impaired memory span: Does comprehension recover? Cortex, 29, 153-159.
3.91. WILSON, B.A., BADDELEY, A.D., SHIEL, A. & Patton, G. (1992). How does post-traumatic amnesia differ from the amnesic syndrome and from chronic memory impairment? Neuropsychological Rehabilitation, 2, 231-243.
3.92. YOUNG, A.W., Flude, B.M., Hellawell, D.J. & Ellis, A.W. (1994a). The nature of semantic priming effects in the recognition of familiar people. British Journal of Psychology, 85.
3.93. YOUNG, A.W., Humphreys, G.W., Riddoch, M.J., Hellawell, D.J. & de Haan, E.H.F. (1994b). Recognition impairments and face imagery. Neuropsychologia, 32, 693-702.
Submitted
3.94. BADDELEY, A.D., EMSLIE, H., KOLODNY, J. & DUNCAN, J. Random generation and the executive control of working memory. (Manuscript submitted to Journal of Experimental Psychology: Learning, Memory, and Cognition).
3.95. DRITSCHEL, B.H., BEKERIAN, D.A. & TOPLIS, R. Retrieving autobiographical memories to problems. (Manuscript submitted to Memory).
3.96. Hodges, J.R., Graham, N. & PATTERSON, K. Charting the progression in semantic dementia: Implications for the organisation of semantic memory. (Manuscript submitted to Memory)
3.97. Hodges, J.R. & PATTERSON, K. Is semantic memory consistently impaired early in the course of Alzheimer's disease? Neuroanatomical and diagnostic implications. (Manuscript submitted to Neuropsychologia).
3.98. Morris, R.G., Abrahams, S., BADDELEY, A.D. & Polkey, C.E. Doors and People: Visual and verbal memory following unilateral temporal lobectomy. (Manuscript submitted to Neuropsychologia).
3.99. Russo, R. & ANDRADE, J. The directed forgetting effect in word fragment completion: An application of the process dissociation procedure. (Manuscript submitted to Quarterly Journal of Experimental Psychology).
3.100. SELLEN, A.J., Louie, G., Harris, J.E., & WILKINS, A. What makes intentions come to mind? An in situ study of prospective memory. Submitted to Memory & Cognition.
3.101. Shaw, G., BEKERIAN, D.A. & McCubbin, J. Effects of video violence on hypermnesia for imaginally encoded concrete and abstract words. (Manuscript submitted to Memory)
Invited Chapters and Commentaries
3.102. ANDRADE, J. (in press). Is learning during anaesthesia implicit?: Commentary on Shanks & St. John. Behavioral and Brain Sciences.
3.103. BADDELEY, A.D. (1990). The development of the concept of working memory: Implications and contributions of neuropsychology. In G. Vallar & T. Shallice (Eds.), Neuropsychological Impairments of Short-Term Memory (pp. 54-73). Cambridge: Cambridge University Press.
3.104. BADDELEY, A.D. (1991). Human learning and memory. In A. Öhman & B. Öhngren (Eds.), Two Faces of Swedish Psychology: Frontiers in Perception and Cognition. An Evaluation of Swedish Research in Cognitive Psychology (pp. 133-142). Uppsala, Swedish Science Press.
3.105. BADDELEY, A.D. (1992). Cognitive function and whipworm infection. Parasitology Today, 8, 394-295.
3.106. BADDELEY, A.D. (1992). Consciousness and working memory. Consciousness and Cognition, 1, 3-6.
3.107. BADDELEY, A.D. (1992). Implicit memory and errorless learning: A link between cognitive theory and neuropsychological rehabilitation? In L.R. Squire & N. Butters (Eds.), Neuropsychology of Memory (2nd Edition) (pp. 309-314). New York: Guilford Press.
3.108. BADDELEY, A.D. (1992). Memory theory and memory therapy. In B.A. Wilson & N. Moffat (Eds.), Clinical Management of Memory Problems (Second edition) (pp. 1-31). London: Chapman & Hall.
3.109. BADDELEY, A.D. (1992). What is autobiographical memory? In M.A. Conway, D.C. Rubin, H. Spinnler & W.A. Wagenaar (Eds.), Theoretical Perspectives on Autobiographical Memory (pp. 13-29). The Netherlands: Kluwer Academic Publishers.
3.110. BADDELEY, A.D. (1993). Holy war or wholly unnecessary? Some thoughts on the "conflict" between laboratory studies and everyday memory. In G.M. Davies & R.H. Logie (Eds.), Memory in Everyday Life (pp. 532-536). Amsterdam: Elsevier Science Publishers B.V.
3.111. BADDELEY, A.D. (1993). Working memory and conscious awareness. In A. Collins, S. Gathercole, M. Conway & P. Morris (Eds.), Theories of Memory (pp. 11-28). Hove: Lawrence Erlbaum Associates.
3.112. BADDELEY, A.D. (1993). Working memory or working attention? In A. Baddeley & L. Weiskrantz (Eds.), Attention: Selection, Awareness and Control: A Tribute to Donald Broadbent (pp. 152-170). Oxford: Clarendon Press.
3.113. BADDELEY, A.D. (in press). Is memory all talk? A comment on Edwards, Middleton and Potter. The Psychologist.
3.114. BADDELEY, A.D. (in press). Models of learning and the nature of self: The remembered self and the enacted self: Comments on Wagenaar and on Ross and Buehler. In R. Fivush & U. Neisser (Eds.), The Remembered Self: Sixth Emory Cognition Symposium.
3.115. BADDELEY, A.D. (in press). Working memory: The interface between memory and cognition. In D.L. Schacter & E. Tulving (Eds.), Memory Systems 1994. Cambridge, MA: MIT Press.
3.116. BADDELEY, A., Bressi, S., Della Sala, S., Logie, R. & Spinnler, H. (1990). Deficit progressivo della working memory in dementi di tipo Alzheimer. In D. Salmaso & P. Caffarra (Eds), Normalita E Patologia Delle Funzioni Cognitive Nell'Invecchiamento (pp. 124-131). Franco Angeli.
3.117. BADDELEY, A.D. & Gathercole, S. (1992). Learning to read: The role of the phonological loop. In J. Alegria, D. Holender, J. Junça de Morais & M. Radeau (Eds.), Analytic Approaches to Human Cognition (pp. 153-167). Amsterdam: Elsevier Science Publishers B.V.
3.118. BADDELEY, A.D. & Hitch, G.J. (1993). Working memory. In P.E. Morris & M.A. Conway (Eds.), The Psychology of Memory, Vol. II (pp. 134-176). Aldershot: Edward Elgar Publishing Company. [Reprinted from 1974]
3.119. BADDELEY, A.D. & Logie, R.H. (1992). Auditory imagery and working memory. In D. Reisberg (Ed.), Auditory Imagery (pp. 179-197). Hillsdale, N.J.: Lawrence Erlbaum Associates.
3.120. BADDELEY, A.D. & Warrington, E.K. (1993). Memory coding and amnesia. In P.E. Morris & M.A. Conway (Eds.), The Psychology of Memory, Vol. II (pp. 417-421). Aldershot: Edward Elgar Publishing Company. [Reprinted from 1973]
3.121. BADDELEY, A.D., PAPAGNO, C. & NORRIS, D. (1991). Phonological memory and serial order: A sandwich for TODAM. In W.E. Hockley & S. Lewandowsky (Eds.), Relating Theory and Data: Essays on Human Memory in Honor of Bennet B. Murdock (pp. 175-194). Hillsdale, N.J.: Lawrence Erlbaum Associates.
3.122. BADDELEY, A.D., Prinz, W., Smith, P.T. & Wagenaar, W.A. (1991). Reflections on the organisation of cognitive psychology in Sweden. In A. Öhman & B. Öhngren (Eds.), Two Faces of Swedish Psychology: Frontiers in Perception and Cognition. An Evaluation of Swedish Research in Cognitive Psychology (pp. 161-168). Uppsala, Swedish Science Press.
3.123. BADDELEY, A.D., Thomson, N. & Buchanan, M. (1993). Word length and the structure of short-term memory. In P.E. Morris & M.A. Conway (Eds.), The Psychology of Memory, Vol. II (pp. 177-191). Aldershot: Edward Elgar Publishing Company. [Reprinted from 1975]
3.124. BADDELEY, A.D., Thornton, A., Chua, S.E. & McKenna, P. (in press). Schizophrenic delusions and the construction of autobiographical memory. In D.C. Rubin (Ed.), Constructing our Past: Autobiographical Memory. New York: Cambridge University Press.
3.125. BEKERIAN, D.A. & DENNETT, J.L. (1992). The truth in content analyses of a child's testimony. In F. Lösel, D. Bender & T. Bliesner (Eds.), Psychology & Law - International Perspectives (pp. 335-344). Berlin: Walter de Gruyte.
3.126. BEKERIAN, D.A. & DENNETT, J.L. (in press). An introduction to the cognitive interview technique. In T. Ney (Ed.), Allegations in Child Sexual Abuse: Assessment and Management. New York: Brunner/Mazel.
3.127. BEKERIAN, D.A. & DENNETT, J.L. (in press). Assessing the truth in children's statements. In T. Ney (Ed.), Allegations in Child Sexual Abuse: Assessment and Management. New York: Brunner/Mazel.
3.128. BEKERIAN, D.A. & DENNETT, J.L. (in press). Interview profiles: Establishing how people interview. In G. Davies, S. Lloyd-Bostock, M. McMuran & C. Wilson (Eds.), Psychology and Law: Recent Advances in Research. Berlin: de Gruyter.
3.129. BEKERIAN, D.A., DENNETT, J.L., Hill, K. & Hitchcock, R. (1992). Effects of detailed imagery on simulated witness recall. In F. Lösel, D. Bender & T. Bliesner (Eds.), Psychology & Law - International Perspectives (pp. 302-308). Berlin: Walter de Gruyte.
3.130. BEKERIAN, D.A. & DRITSCHEL, B.H. (1992). Autobiographical remembering: An integrated approach. In M. Conway, D. Rubin, H. Spinnler & W. Wagenaar (Eds.), Theoretical Perspectives on Autobiographical Memory (pp. 135-150). The Netherlands: Kluwer Academic Publishers.
3.131. Dickinson, A. & SHANKS, D.R. (1994). Instrumental action and causal representation. In G. Lewis, D. Premack & D. Sperber (Eds.), Causal Understandings in Cognition and Culture. Oxford: Oxford University Press.
3.132. Gathercole, S.E. & BADDELEY, A.D. (1993). Evaluation of the role of phonological STM in the development of vocabulary in children: A longitudinal study. In P.E. Morris & M.A. Conway (Eds.), The Psychology of Memory, Vol. II (pp. 192-205). Aldershot: Edward Elgar Publishing Company. [Reprinted from 1989]
3.133. HOUGHTON, G. (1990). The problem of serial order: A neural network model of sequence learning and recall. In R. Dale, C. Mellish & M. Zock (Eds.), Current Research in Natural Language Generation (pp. 287-319). London: Academic Press.
3.134. Lachnit, H., Kimmel, H., Bevill, M., Martin, I., LEVEY, A. & Hamm, A. (1990). Classical conditioning with human subjects. In P.D. Drenth, J.A. Sergeant & R.J. Takens (Eds.), European Perspective in Psychology, Vol. 1, Section V, Cognitive Psychology (pp. 353-368). Chichester: John Wiley & Sons.
3.135. Logie, R.H. & BADDELEY, A.D. (1990). Imagery and working memory. In P.J. Hampson, D.F. Marks & J. Richardson (Eds.), Imagery: Current Developments (pp. 103-128). London: Routledge.
3.136. MAYLOR, E.A. (in press a). Does prospective memory decline with age? In M. Brandimonte, G. Einstein & M. McDaniel (Eds.), Prospective Memory: Theory and Applications. Hillsdale, N.J.: Lawrence Erlbaum Associates.
3.137. McKenna, P. & BADDELEY, A.D. (in press). Memory in schizophrenia. In R. Campbell & M. Conway (Eds.), Broken Memories. Oxford: Blackwell Publishers.
3.138. MURRE, J.M.J. (in press). Connectionism and other approaches to natural computation: An overview. In T. Dijkstra & K. De Smedt (Eds.), Computational Psycholinguistics: Symbolic and Subsymbolic Models of Language Processing. Hemel Hempstead: Harvester Wheatsheaf.
3.139. MURRE, J.M.J. (in press). Neurosimulators. In M.A. Arbib (Ed.), Handbook of Brain Research and Neural Networks. Cambridge, MA: MIT Press.
3.140. PATTERSON, K. & Hodges, J. (in press). Disorders of semantic memory. In A.D. Baddeley, B.A. Wilson & F.N. Watts (Eds.), Handbook of Memory Disorders. Chichester, Sussex: John Wiley & Sons.
3.141. PATTERSON, K. (1991). Learning by association: Two tributes to George Mandler. In W. Kessen, A. Ortony & F. Craik (Eds.), Memories, Thoughts and Emotions: Essays in Honour of George Mandler (pp. 35-41). Hillsdale, N.J.: Lawrence Erlbaum Associates.
Conference Proceedings
3.142. BADDELEY, A.D. (1992). Implications of neuropsychological evidence for theories of normal memory. In Proceedings of International School of Neuroscience course on Neuropsychology: The Neuronal Basis of Cognitive Function, Vol. 2 (pp. 91-101). New York: Thieme Medical Publishers Inc.
3.143. BADDELEY, A.D., Gathercole, S. & PAPAGNO, C. (1991). Short-term phonological memory: A crucial system for language development? In Current Issues in Natural Language Processing. (Proceedings of a Conference held at University of Texas, Feb. 15-16, 1991), Center for Cognitive Science, University of Texas.
3.144. MURRE, J.M.J. (1992). The effects of pattern presentation on interference in backpropagation networks. In Proceedings of the Fourteenth Annual Conference of the Cognitive Science Society (pp. 54-59). Hillsdale, NJ: Lawrence Erlbaum.
3.145. SHANKS, D.R. (1991). Some parallels between associative learning and object classification. In J.-A. Meyer & S.W. Wilson (Eds.), From Animals to Animats: Proceedings of the First International Conference on Simulation and Adaptive Behavior (pp. 337-343). Cambridge, MA: MIT Press.
Technical Reports and Theses
3.146. BEKERIAN, D.A. & DENNETT, J.L. (1990). Interview Profiles (Parts 1 and 2). Child Sexual Abuse, Joint Investigative Project, Cambs. Police and Social Services.
3.147. BEKERIAN, D.A. & DENNETT, J.L. (1990). Interview Profiles (Part 3). Summaries, Profiles and Final Report. Child Sexual Abuse, Joint Investigative Project, Cambs. Police and Social Services.
3.148. Eldridge, M., SELLEN, A.J. & BEKERIAN, D. (1992). Memory problems at work: Their range frequency, and severity. EuroPARC Technical Report No. EPC-92-103.
3.149. GREEN, R.E.A. (1992). Investigations of intentional and automatic processing in amnesic, healthy elderly and healthy young subjects. Unpublished PhD Thesis, University of Cambridge.
3.150. KOLODNY, J. (1993). Conscious and unconscious processes in learning and memory retrieval. Unpublished PhD Thesis, University of Cambridge.
3.151. SHANKS, D.R. & Gluck, M.A. (1991). Tests of an adaptive network model for the identification, categorization, and recognition of continuous-dimension stimuli. Technical Report No. 9103, Department of Cognitive Science, University of California, San Diego, La Jolla, CA.
3.152. SHANKS, D.R. & St John, M.F. (1992). Characteristics of dissociable human learning systems. Technical Report No. 9203, Department of Cognitive Science, University of California, San Diego, La Jolla, CA.
Tests and Patents
3.153. BADDELEY, A.D., EMSLIE, H. & NIMMO-SMITH, I. (1992). The Speed and Capacity of Language Processing (SCOLP) Test. Bury St Edmunds, Suffolk: Thames Valley Test Company.
3.154. BADDELEY, A.D., EMSLIE, H. & NIMMO-SMITH, I. (in press). Doors and People: A test of visual and verbal recall and recognition. Flempton, Bury St Edmunds: Thames Valley Test Company.
3.155. Kopelman, M., WILSON, B. & BADDELEY, A.D. (1990). The Autobiographical Memory Interview. Thames Valley Test Company.
3.156. Howard, D. & PATTERSON, K. (1992). Pyramids and Palmtrees: A test of semantic access from words and pictures. Bury St Edmunds, Suffolk: Thames Valley Test Company.
Dissemination
3.157. ANDRADE, J. (1993). Consciousness: Current views. In J.G. Jones (Ed.), Depth of Anesthesia: International Anesthesiology Clinics, Vol. 31 (pp. 13-25). Boston: Little, Brown and Co.
3.158. ANDRADE, J. & BADDELEY, A.D. (1993). Human memory and anaesthesia. In J.G. Jones (Ed.), Depth of Anesthesia: International Anesthesiology Clinics, 31 (pp. 39-51). Boston: Little, Brown and Co.
3.159. BADDELEY, A.D. (1990). Acoustic memory and language. Current Contents, 20, 24.
3.160. BADDELEY, A.D. (1990). The MRC Applied Psychology Unit. MRC News, June 1990, Issue No. 47, 14-17.
3.161. BADDELEY, A.D. (1993). Theories of normal memory. In F.J. Stachowiak, R. De Bleser, G. Deloche, R. Kaschel, H. Kremin, P. North, L. Pizzamiglio, I. Robertson, & B. Wilson, (Eds.), Developments in the Assessment and Rehabilitation of Brain-damaged Patients: Perspectives from a European Concerted Action (pp. 93-97). Tübingen: Gunter Narr Verlag.
3.162. BADDELEY, A.D. (1993). Verbal and visual subsystems of working memory. Current Biology, 3, 563-565.
3.163. BADDELEY, A.D. (1994). Les memoires humaines. La Recherche, 25, 730-735.
3.164. BADDELEY, A.D. (1994). Memory. In A.M. Colman (Ed.), Companion Encyclopedia of Psychology, Vol. 1 (pp. 281-301). London: Routledge.
3.165. BADDELEY, A.D. (in press). The psychology of memory. In A. D. Baddeley, B. A. Wilson, & F. N. Watts (Eds.), Handbook of Memory Disorders. Chichester: John Wiley.
3.166. BADDELEY, A.D. (in press). Working memory. In M. Gazzaniga (Ed.), Handbook of Cognitive Neuroscience. Cambridge, Mass: MIT Press.
3.167. BADDELEY, A.D. (in press). Working memory. In L. Squire (Ed.), The Encyclopedia of Learning and Memory. New York: Macmillan Publishing Company.
3.168. BEKERIAN, D.A. & DENNETT, J.L. (1990). To gain reliable evidence, child abuse victims need to be questioned carefully (Cambridgeshire project which looked at interviewing styles and techniques). In Community Care (Supplement Child Care), Nov. issue, p.7.
3.169. BEKERIAN, D.A. & DENNETT, J.L. (1991). Interviewing abused children. Policing, 7, 355-360.
3.170. McKenna, P. J., CLARE, L., & BADDELEY, A. D. (in press). Schizophrenia. In A. D. Baddeley, B. A. Wilson, & F. N. Watts (Eds.), Handbook of Memory Disorders. Chichester: John Wiley.
3.171. PATTERSON, K. (1991). Cognitive neuropsychology. Medical Research Council Annual Report, 1990/1991 (pp. 28-33). London: MRC.
3.172. PATTERSON, K. & Hodges, J.R. (1992). Progressive deterioration of semantic memory and language. MRC News, June, No. 55, 6-7.
3.173. SHANKS, D.R. (1991). Remembrance of things unconscious. New Scientist, 131, (No. 1783), 39-42.
3.174. SHANKS, D.R. (1993). Breaking Chomsky's rules. New Scientist, 138, (No. 1858), 26-30.
REFERENCES TO OTHER WORK
BADDELEY, A. (1968). How does acoustic similarity influence short-term memory? Quarterly Journal of Experimental Psychology, 20, 249-264.
BADDELEY, A. (1986). Working Memory. Oxford: Oxford University Press.
BARNARD, P.J. & TEASDALE, J.D. (1991). Interacting cognitive subsystems: A systemic approach to cognitive-affective interaction and change. Cognition and Emotion, 5, 1-39.
Bermùdez, J., MARCEL, A.J. & Eilan, N. (Eds.) (in press). The Body and the Self. Bradford Books: MIT Press.
Braak, H. & Braak, E. (1991). Neuropathological staging of Alzheimer-related changes. Acta Neuropathologica, 82, 239-259.
Brédart, S. (1993). Retrieval failures in face naming. Memory, 1, 351-366.
Bruce, V. & Valentine, T. (1985). Identity priming in the recognition of familiar faces. British Journal of Psychology, 76, 363-383.
Bruce, V. & Valentine, T. (1986). Semantic priming of familiar faces. Quarterly Journal of Experimental Psychology, 38A, 125-150.
Bruce, V. & YOUNG, A. (1986). Understanding face recognition. British Journal of Psychology, 77, 305-327.
Burgess, N. & Hitch, G., (1992). Towards a network model of the articulatory loop. Journal of Memory and Language, 31, 429-460.
Burton, A.M. (in press). Learning new faces in an interactive activation and competition model. Visual Cognition.
Burton, A.M. & Bruce, V. (1993). Naming faces and naming names: exploring an interactive activation model of person recognition. Memory, 1, 457-480.
Burton, A.M., Bruce, V. & Johnston, R.A. (1990). Understanding face recognition with an interactive activation model. British Journal of Psychology, 81, 361-380.
Burton, A.M., YOUNG, A.W., Bruce, V., Johnston, R. & Ellis, A.W. (1991). Understanding covert recognition. Cognition, 39, 129-166.
de Haan, E.H.F., YOUNG, A. & Newcombe, F. (1987). Face recognition without awareness. Cognitive Neuropsychology, 4, 385-415.
de Haan, E.H.F., YOUNG, A.W. & Newcombe, F. (1991a). Covert and overt recognition in prosopagnosia. Brain, 114, 2575-2591.
de Haan, E.H.F., YOUNG, A.W. & Newcombe, F. (1991b). A dissociation between the sense of familiarity and access to semantic information concerning familiar people. European Journal of Cognitive Psychology, 3, 51-67.
De Renzi, E., Faglioni, P., Grossi, D. & Nichelli, P. (1991). Apperceptive and associative forms of prosopagnosia. Cortex, 27, 213-221.
Ellis, A.W. (1992). Cognitive mechanisms of face processing. Philosophical Transactions of the Royal Society, London, B335, 113-119.
Ellis, A.W., YOUNG, A.W. & Critchley, E.M.R. (1989). Loss of memory for people following temporal lobe damage. Brain, 112, 1469-1483.
Flude, B.M., Ellis, A.W. & Kay, J. (1989). Face processing and name retrieval in an anomic aphasic: names are stored separately from semantic information about familiar people. Brain and Cognition, 11, 60-72.
Funnell, E. (1993). Breakdown of object concepts and the organisation of semantic memory. Paper resented to the Experimental Psychology Society, Cambridge, July 1993.
Hanley, J.R., Pearson, N. & YOUNG, A.W. (1990). Impaired memory for new visual forms. Brain, 113, 1131-1148.
Hanley, J.R., YOUNG, A.W. & Pearson, N. (1989). Defective recognition of familiar people. Cognitive Neuropsychology, 6, 179-210.
Hanley, J.R., YOUNG, A.W. & Pearson, N. (1991). Impairment of the visuo-spatial sketch pad. Quarterly Journal of Experimental Psychology, 43A, 101-125.
Hay, D.C., YOUNG, A.W. & Ellis, A.W. (1991). Routes through the face recognition system. Quarterly Journal of Experimental Psychology, 43A, 761-791.
Hodges, J.R. (1993). Pick's disease. In A. Burns & R. Levy (Eds.), Dementia. London: Chapman and Hall.
Hodges, J.R., Salmon, D.P. & Butters, N. (1992). Semantic memory impairment in Alzheimer's disease: Failure of access or degraded knowledge? Neuropsychologia, 30, 301-314.
Humphreys, G.W., Troscianko, T., Riddoch, M.J., Boucart, M., Donnelly, N. & Harding, G.F.A. (1992). Covert processing in different visual recognition systems. In A.D. Milner & M.D. Rugg (Eds.), The Neuropsychology of Consciousness (pp. 39-68). London: Academic Press.
Kopelman, M.D., WILSON, B.A. & BADDELEY, A.D. (1989). The autobiographical memory interview: A new assessment of autobiographical and personal semantic memory in amnesic patients. Journal of Clinical and Experimental Neuropsychology, 11, 724-744.
Lamming, M. & Newman, W. (1992). Activity-based information retrieval: Technology in support of personal memory. In F.H. Vogt (Ed.), Information Processing '92: Proceedings of the 12th World Computer Congress , Vol. III. (pp. 68-81). Amsterdam: Elsevier Science Publishers.
Levine, D.N. & Calvanio, R. (1989). Prosopagnosia: A defect in visual configural processing. Brain and Cognition, 10, 149-170.
Mandler, J. (1990). Recall of events by preverbal children. In A. Diamond (Ed.), The Development and Neural Bases of Higher Cognitive Functions. New York: New York Academy of Science.
Mandler, J., Bauer, P.J. & McDonough, L. (1991). Separating the sheep from the goats: Differentiating global categories. Cognitive Psychology, 23, 263-298.
Marr, D. (1982). Vision. San Francisco: Freeman.
MAYLOR, E.A. (1990a). Age and prospective memory. Quarterly Journal of Experimental Psychology, 42A, 471-493.
MAYLOR, E.A. (1990b). Age, blocking and the tip of the tongue state. British Journal of Psychology, 81, 123-134.
MAYLOR, E.A. (1990c). Recognizing and naming faces: Aging, memory retrieval and the tip of the tongue state. Journal of Gerontology: Psychological Sciences, 45, 215-226.
MAYLOR, E.A. (1991). Recognizing and naming tunes: Memory impairment in the elderly. Journal of Gerontology: Psychological Sciences, 46, 207-217.
MAYLOR, E.A. (1993). Minimized prospective memory loss in old age. In J. Cerella, J. Rybash, W. Hoyer & M.L. Commons (Eds.), Adult Information Processing: Limits on Loss (pp. 529-551). San Diego: Academic Press.
MAYLOR, E.A. & Rabbitt, P.M.A. (1993). Alcohol, reaction time and memory: A meta-analysis. British Journal of Psychology, 84, 301-317.
MAYLOR, E.A., Rabbitt, P.M.A., James, G.H. & Kerr, S.A. (1992). Effects of alcohol, practice and task complexity on reaction time distributions. Quarterly Journal of Experimental Psychology, 44A, 119-139.
MAYLOR, E.A. & Valentine, T. (1992). Linear and nonlinear effects of aging on categorizing and naming faces. Psychology and Aging, 7, 317-323.
McCarthy, R. (1994). Semantic quadrants: Evidence for multiple knowledge bases. Paper presented to the British Neuropsychological Society, London, March 1994.
McClelland, J.L. & Rumelhart, D.E. (1988). Explorations in Parallel Distributed Processing. Cambridge, Massachusetts: Bradford Books.
McNeil, J.E. & Warrington, E.K. (1991). Prosopagnosia: A reclassification. Quarterly Journal of Experimental Psychology, 43A, 267-287.
Morrison, C.M., Ellis, A.W. & Quinlan, P.T. (1992). Age of acquisition, not word frequency, affects object naming, not object recognition. Memory and Cognition, 20, 705-714.
MURRE, J.M.J. (1992). Categorisation and Learning in Modular Neural Networks. Hemel Hempstead: Harvester Wheatsheaf. Co-published by Lawrence Erlbaum in the USA and Canada (Hillsdale, NJ).
Rabbitt, P.M.A. & MAYLOR, E.A. (1991). Investigating models of human performance. British Journal of Psychology, 82, 259-290.
Rapp, B.A. & Caramazza, A. (1993). On the distinction between deficits of access and deficits of storage: A question of theory. Cognitive Neuropsychology, 10, 113-141.
Ross, E.D. (1980). Sensory-specific and fractional disorders of recent memory in man, 1. Isolated loss of visual recent memory. Archives of Neurology, 37, 193-200.
Schacter, D.L. (1987). Implicit memory: history and current status. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 501-518.
SHALLICE, T. (1988). From Neuropsychology to Mental Structure. Cambridge: Cambridge University Press.
Schneider, W. & Detweiler, M. (1987). A connectionist/control architecture for working memory. In G.H. Bower (Ed.), The Psychology of Learning and Motivation, Vol. 21. New York: Academic Press.
Snowden, J.S., Neary, D., Mann, D.M.A., Goulding, P.J. & Testa, H.J. (1992). Progressive language disorder due to lobar atrophy. Annals of Neurology, 31, 174-183.
Tranel, D. & Damasio, A.R. (1985). Knowledge without awareness: An autonomic index of facial recognition by prosopagnosics. Science, 228, 1453-1454.
Valentine, T., Moore, V., Flude, B.M., YOUNG, A.W. & Ellis, A.W. (1993). Repetition priming and proper name processing. Do common names and proper names prime each other? Memory, 1, 329-349.
Warrington, E.K. (1984). Recognition Memory Test. Windsor: NFER-Nelson.
YOUNG, A.W. (1992). Face recognition impairments. Philosophical Transactions of the Royal Society, London, B335, 47-54.
YOUNG, A.W. (1993). Recognising friends and acquaintances. In G.M. Davies & R.H. Logie (Eds.), Memory in Everyday Life (pp. 325-350). Amsterdam: North Holland.
YOUNG, A.W. (1994a). Conscious and nonconscious recognition of familiar faces. In C. Umiltà & M. Moscovitch (Eds.), Attention and Performance, XV: Conscious and Nonconscious Information Processing (pp. 153-178). Cambridge, Massachusetts: MIT Press/Bradford Books.
YOUNG, A.W. (1994b). Face recognition. In G. d'Ydewalle, P. Eelen and P. Bertelson (Eds.), International Perspectives on Psychological Science, volume 2: The State of the Art (pp. 1-27). Hove, East Sussex: Lawrence Erlbaum.
YOUNG, A.W. (1994c). Neuropsychology of awareness. In A. Revonsuo & M. Kamppinen (Eds.), Consciousness in Philosophy and Cognitive Neuroscience (pp. 173-203). Hillsdale, New Jersey: Lawrence Erlbaum.
YOUNG, A.W. & Bruce, V. (1991). Perceptual categories and the computation of 'grandmother'. European Journal of Cognitive Psychology, 3, 5-49.
YOUNG, A.W. & de Haan, E.H.F. (1992). Face recognition and awareness after brain injury. In A.D. Milner & M.D. Rugg (Eds.), The Neuropsychology of Consciousness (pp. 69-90). London: Academic Press.
YOUNG, A.W., Hay, D.C. & Ellis, A.W. (1985). The faces that launched a thousand slips: Everyday difficulties and errors in recognizing people. British Journal of Psychology, 76, 495-523.
Collaborations
Baddeley
UK based
Della Sala - Psychology, Aberdeen
Gathercole - Psychology, Bristol
Hodges - Neurology, Cambridge
Jones - Anaesthetics, Cambridge
Logie - Psychology, Aberdeen
McKenna - Psychiatry, Cambridge
Morris - Institute of Psychiatry, London
Nicolson - Psychology, Sheffield
Robbins - Psychology, Cambridge
Outside UK
Becker - Psychiatry, Pittsburgh
McGregor - Nutrition, U West Indies
Papagno - Neurology, Milan
Salamé - Psychiatry, Strasbourg
Spinnler - Neurology, Milan
White - Psychology, Otago
Bekerian
UK based
Eldridge - Rank Xerox, Cambridge
Dritschel - Psychology, University of East London
Clifford, Toplis - Psychology, University of East London
Davies - Psychology, Leicester
Outside UK
Shaw - Psychology, Georgetown College, Kentucky
McCubbin - Psychology, Kentucky
Jackson - NISCALE, The Netherlands
Undeutsch - Psychology, Köln
Foa - Psychology, Pennsylvania
Maylor
UK based
Conway - Psychology, Bristol
Rabbit - Psychology, Manchester
Ellis - Psychology, Reading
Murre
UK based
Broeder - Psychology, Tilburg
Outside UK
Happel, Heemskerk, Hudson - Psychology, Leiden
K Patterson
UK based
Hodges - Neurology, Cambridge
A Young
UK based
Humphreys, Riddoch, Psychology, Birmingham
Newcombe - Psychology, Oxford
Ellis - Psychology, York
Flude - Psychology, Lancaster
Calder - Psychology, Durham
Outside UK
de Haan - Psychology, Utrecht